Assessing accessibility: The effects of accessibility ...Canadian cities, however, have been slow to...
Transcript of Assessing accessibility: The effects of accessibility ...Canadian cities, however, have been slow to...
ASSESSING ACCESSIBILITY: THE EFFECTS
OF ACCESSIBILITY IMPROVEMENTS ON
NEIGHBOURHOOD DEVELOPMENT
SUPERVISED RESEARCH PROJECT | ROBBIN DEBOOSERE
Assessing accessibility: The effects of accessibility improvements on
neighbourhood development
Prepared by Robbin Deboosere
Supervised by Ahmed El-Geneidy
School of Urban Planning McGill University
Montréal, Québec, Canada
April 2018
Supervised research project Submitted in partial fulfillment of the requirements of the degree of Master of Urban Planning
ASSESSING ACCESSIBILITY: THE EFFECTS OF ACCESSIBILITY IMPROVEMENTS ON NEIGHBOURHOOD DEVELOPMENT
Transport and land use are inextricably linked. Growth occurs where transport infrastructure is
constructed, and transport investments are made where land use is changing. Yet current planning documents tend to isolate transport by solely focusing on mobility, or how fast we can travel, while disregarding the destinations we want to reach.
To combat this flawed planning paradigm, accessibility indicators are increasingly being used in both research and practice. Accessibility improves upon the concept of mobility by also considering potential and desirable destinations, instead of only examining our ability to move through the transport network. The most common accessibility indicator measures the number of jobs that can be reached within a certain time threshold. Accessibility thus recognizes that most of us travel not for its own sake, but to reach a destination.
While cities such as London, Paris, and Sydney are now employing the concept in their transport plans, Canadian cities are lagging far behind and are missing out on the myriad of benefits unlocked by accessibility planning. To quantify these benefits, this study has examined both the social and economic effects of accessibility improvements on neighbourhood development.
Neighbourhood benefits of accessibility
Apart from the direct transport benefits, policies aimed at improving accessibility can help Canada’s neighbourhoods in the following five ways:
1. Increases in accessibility are related to increases in household income. The most vulnerable census tracts in Toronto where accessibility improved saw their income increase by $3,000 more than similar areas where accessibility remained stable.†
2. Accessibility improvements are associated with relative decreases in unemployment rates. The most vulnerable census tracts in Toronto where accessibility improved saw their unemployment rate rise by 1% less than similar areas where accessibility remained stable.†
3. Higher accessibility to jobs results in shorter commutes. If the average accessibility level in Toronto would be doubled, commute times would decrease by 2.12 minutes.
4. High accessibility locations attract more residential, commercial, and industrial development. If average accessibility to workers in Toronto would be doubled, each neighbourhood would attract 44 extra jobs per square kilometer.
5. Transit modal share is higher in high accessibility neighbourhoods. For every 100,000 jobs accessible by public transport, transit modal share increases by 4.25% (up to 6.21% in socially vulnerable areas).
What should policy makers do?
Here is how Canadian cities can seize the opportunity to be the world leaders in innovative accessibility planning:
1. Incorporate accessibility into transport planning documents. By using explicit accessibility objectives to reach planning goals, each decision will be based on a comprehensive understanding of the transport and land use interaction.
2. Evaluate transport investments through accessibility. Consider which areas will see their accessibility levels improve from the investment, and thereby also their income, unemployment rates, commutes, economic development, and transit mode share. Planners can thus use accessibility to foster socially inclusive environments that are also conducive to development.
3. Monitor progress of key planning goals and objectives through accessibility indicators.
Through these recommendations, Canadian cities can move away from the outdated mobility planning paradigm and instead utilize accessibility-oriented development techniques, and thereby maximize the benefits resulting from their transport investments.
† For every one unit increase in a competitive accessibility metric, which takes into account competition for jobs
III
ACKNOWLEDGEMENTS
I would first like to thank Ahmed El-Geneidy, my supervisor for this project, for his guidance and
support throughout my research and my Master’s degree. I would also like to thank my second
reader, Geneviève Boisjoly, for her thoughtful feedback and her contributions as co-author of the
second chapter of this supervised research project.
I would also like to thank David Levinson for his contributions as co-author of the third chapter of
this supervised research project. A thank you is also in order to David King, Nicholas Day and
Joshua Engel-Yan from Metrolinx for providing access to their data.
Finally, I would like to express my thanks to the Social Sciences and Humanities Research Council
of Canada and the Natural Sciences and Engineering Research Council of Canada for funding
this research.
IV
TABLE OF CONTENTS
Introduction 1
Chapter 1: Evaluating accessibility to low-income jobs across Canada 4
1. Literature review 6
2. Data and methodology 9
3. Results and Discussion 11
4. Conclusion 18
Chapter 2: Understanding the relationship between changes in accessibility
to jobs, income and unemployment in Toronto, Canada 20
1. Equity of accessibility and equity of outcome 21
1.1 Accessibility 21
1.2 Equity of accessibility 22
1.3 Accessibility, unemployment and income 23
2. Data and methodology 24
2.1 Study context 24
2.2 Data 25
2.3 Methodology 26
3. Results and discussion 27
3.1 Vertical equity 31
3.2 Linear regression models 32
4. Conclusion 35
Chapter 3: Accessibility-oriented development 37
1. Literature review 39
1.1 Accessibility 39
1.2 Transport, accessibility and economic development 40
2. Hypotheses, data, and methodology 42
2.1 Hypotheses 42
2.2 Case study 44
2.3 Data and methods 45
3. Results and discussion 48
3.1 Accessibility, commute duration and urban development 50
4. Conclusion 54
Conclusion 56
V
LIST OF FIGURES
Figure 1 - Overview of the 11 metropolitan areas included in the study 6
Figure 2 - Accessibility to all jobs in Toronto, Montreal and Vancouver 14
Figure 3 - Accessibility to low-income jobs in Toronto, Montreal and Vancouver 15
Figure 4 - Comparison of accessibility to all jobs and accessibility to low-income jobs 16
Figure 5 - Accessibility and public transport mode share 18
Figure 6 - Context map 26
Figure 7 - Median household income and unemployment rate in the GTHA in 2001 and 2011 29
Figure 8 - Transit accessibility in the GTHA in 2001 and 2011 30
Figure 9 - Relative competitive accessibility by transit, by income decile in the GTHA 31
Figure 10 - Context map of the Greater Toronto and Hamilton Area 45
Figure 11 - Accessibility to jobs and the labour force by car within 30 minutes 49
Figure 12 - Accessibility to jobs and the labour force by public transport within 45 minutes 50
LIST OF TABLES
Table 1 - Summary statistics for the 11 metropolitan areas 13
Table 2 - Summary statistics 27
Table 3 - Regression results for census tract median household income and unemployment rate in
2011 in the Greater Toronto and Hamilton area 34
Table 4 - Predicted 2011 income and unemployment rates for each income decile in 2001 35
Table 5 - Summary of AOD hypotheses 43
Table 6 - Descriptive statistics for commute duration, accessibility and development data 47
Table 7 - Regression model predicting average commute duration in 2001 51
Table 8 - Commute time and open area in 2011 fitted to accessibility in 2001 and changes in
accessibility between 2001 and 2011 52
Table 9 - Job and population density in 2011 fitted to accessibility in 2001 and changes in
accessibility between 2001 and 2011 54
1
INTRODUCTION
“It is clear that transport is a main factor in all large-scale town-planning. The arrangements of houses
must always have a direct relation to the daily movements of the inhabitants. […] Town-building and
transport are two elements in one problem that ought never to be separated. […] There can be no
good regional planning in London or anywhere else unless the location and the movement of
population are treated as one problem.” (The Spectator, 1933)
The concept of integrated land use and transport planning was not a new idea. In 19th century
London, urban thinkers had already recognized the inherent connection between transport and
land use. In 1846, Charles Pearson advocated the joint development of both railways and housing
estates, while other reformers in Victorian London recognized similar opportunities and
encouraged the construction of railways to disperse London’s ever-growing population (Barker &
Robbins, 1963; Dyos, 1955; Haywood, 1997). Even in the unregulated, laissez-faire climate
enjoyed by railways at the time, residential growth occurred where railways were constructed;
similar mechanisms were at play in the railroad towns of the North American hinterland, the
expansion of Berlin, or Boston’s streetcar suburbs (Ahlfeldt & Wendland, 2011; Sultana, 2017;
Warner, 1962).
Similar calls to action, advocating an integrated view on land use and transport planning, were
issued en masse in the 1960s (see, for example, W. Owen (1966) and Mumford (1968)). Yet
mobility planning, which had gained traction by traffic engineers in the 1950s and 1960s due to
the dominance of the private automobile, was the prevailing paradigm at the time: transport was
isolated from land use (P. W. G. Newman & Kenworthy, 1996). Endless suburban sprawl resulted.
In the last few decades, planners, increasingly concerned with traffic congestion, sprawl and
environmental degradation, have slowly started to shift away from the concept of mobility planning
towards accessibility planning – a shift accelerated by Calthorpe’s transit-oriented developments
and the New Urbanism movement (Calthorpe, 1993; Cervero, 1997; Congress for the New
Urbanism, 1999; P. W. G. Newman & Kenworthy, 1996). The accessibility planning paradigm
2
intends to revalorize the land use and transport connection, while re-focusing planning on people
and away from the automobile (Cervero, 1997).
Accessibility, or the ease of reaching destinations, is now increasingly being used in both research
and practice as a key performance indicator for the accessibility planning paradigm (Boisjoly & El-
Geneidy, 2017b; D. G. Proffitt et al., 2017). Accessibility, usually operationalized as the number
of opportunities (such as jobs or schools) an individual can reach within a certain time threshold,
improves upon the concept of mobility by also considering potential and desirable destinations,
instead of only examining an individual’s ability to move through the transportation network. As
such, accessibility to destinations explicitly considers the connection between land use and
transportation (Geurs & van Wee, 2004; Handy & Niemeier, 1997; Hansen, 1959; Wickstrom,
1971).
Metropolitan transport plans, such as those in London, Paris, Sydney, and Atlanta, are now
increasingly employing the concept of accessibility, either as an independent goal or objective, or
as part of an environmental justice assessment (Atlanta Regional Commission, 2016; Conseil
régional d'Île‐de‐France, 2014; NSW Government, 2012; San Diego Association of Governments,
2011; Transport for London, 2006). Authors of these transport plans argue that increasing access
to opportunities can reduce social disadvantage while at the same time acting as a catalyst for
economic growth. The plan for Sydney, for example, mentions supporting regional development
by improving accessibility to jobs, while in London improved access to employment is used as a
proxy for social inclusion (NSW Government, 2012; Transport for London, 2006).
Yet despite these hypothesised benefits, planners across Canada have been slow to adopt the
concept. Montréal’s 2008 transportation plan mentions the term vaguely, but does not include it
as an objective or goal (Ville de Montréal, 2008). In Toronto, the 2008 Big Move issued by
Metrolinx only provides a crude indicator for accessibility to public transport, and not to potential
destinations (Metrolinx, 2008). Only the 2016 discussion paper includes accessibility to jobs,
services and major destinations as an objective, and provides accessibility maps (Metrolinx,
2016). In contrast, Vancouver’s 2014 transport plan does not even mention accessibility to
destinations (Mayor's Council on Regional Transportation, 2014).
The transport plan for Calgary incorporates accessibility to the transit network, accessibility to daily
needs and accessibility to major destinations, but defines the concept vaguely as the “ease of
access/egress to any location by walking, cycling, transit, and private vehicles, or for commercial
vehicles” (The City of Calgary, 2009). Edmonton’s transport plan mentions accessibility to
3
employment opportunities, but only in the context of complete communities (City of Edmonton,
2009). Similarly, the plan for Ottawa focuses only on access to transit stations and not to
destinations, although some transit priority investments are rationalized through accessibility to
employment (City of Ottawa, 2013). Thus, among major Canadian metropolitan areas, only
Toronto is in the process of adopting accessibility to destinations as a key performance indicator.
In particular, Metrolinx aims to improve access to employment and other opportunities for
vulnerable populations, mirroring similar policies in London and Sydney aimed at creating
economic growth and reducing inequality.
To increase the uptake of accessibility planning among Canadian cities, the benefits of such an
approach should be more carefully examined: what can cities gain from adopting accessibility
planning and indicators? What could be the benefits of policies aimed at improving accessibility
levels? Only when planners are aware of these effects can they start to adopt the concept of
accessibility to guide planning policies in their city.
However, while a large body of literature has assessed accessibility levels for different cities,
socio-economic groups and neighbourhoods, or changes in these accessibility levels over time,
little research has been conducted to assess the outcomes of improvements in accessibility: what
will occur to neighbourhoods of different socio-economic status when accessibility levels change?
Can government policies, such as those in Sydney, London and Toronto, targeting accessibility
improvements in socially deprived neighbourhoods, help residents living in these areas? And can
accessibility improvements provide a catalyst for economic growth?
This supervised research project examines the impacts of changes in accessibility levels on
neighbourhood development. The first chapter provides a comprehensive look at the state of
accessibility in Canada’s largest cities and compares these cities in terms of the equity of their
transport and land use systems. In the second chapter, the social benefits of accessibility
changes are investigated, while the third and final chapter presents a study on the economic
benefits occurring from changes in accessibility levels. Both chapter two and three focus on the
context of the Greater Toronto and Hamilton Area, Canada’s largest metropolis, housing around
seven million inhabitants. Social benefits are measured through changes in neighbourhood
median income and unemployment rates, while economic benefits are measured through
residential and job density.
4
CHAPTER 1: EVALUATING ACCESSIBILITY TO LOW-INCOME JOBS ACROSS CANADA1
Transport and land use planning are inextricably linked. Although the modern concept of
integrated transport and land use planning goes back to at least the 1930s (one could even argue
as far back as 1846 when Charles Pearson suggested coupling housing estates and railway
stations in London), interest seems to have been renewed after Calthorpe’s proposal advocating
transit-oriented developments and the rise of the New Urbanism movement (Barker & Robbins,
1963; Calthorpe, 1993; The Spectator, 1933). Since, transport planners have started to move
away from the dominant mobility planning paradigm, which had gained traction by traffic engineers
in the 1950s and 1960s due to the dominance of the private automobile, towards the concept of
accessibility planning (D. G. Proffitt et al., 2017).
Accessibility, or the ease of reaching destinations, is now being increasingly used to operationalize
notions of integrated land use and transport planning and to act as a performance indicator for
these concepts (Boisjoly & El-Geneidy, 2017b). Indeed, metropolitan transportation plans, such
as those in London, Paris, Sydney and Atlanta, are now employing the concept, either as an
independent goal or objective, or as part of an environmental justice assessment (Atlanta Regional
Commission, 2016; Conseil régional d'Île‐de‐France, 2014; NSW Government, 2012; Transport
for London, 2006).
Starting with Hansen (1959) (and somewhat later Wickstrom (1971), Ingram (1971) and Dalvi and
Martin (1976), among others), a growing body of literature has been continuously refining the
concept of accessibility. By far the most widely used metric for accessibility is the number of jobs
an individual can reach within a set time limit (known as a cumulative opportunity measure). One
of the inherent strengths of this measure is that it makes comparisons between different socio-
1 Co-authored with Ahmed El-Geneidy
5
economic groups fairly straightforward. It is thus not surprising that accessibility, especially in
literature focusing on equity concerns, has often been used to identify which groups are being well
served, and which are being underserved, or to predict who will benefit from a new transport or
land use project. Contemporary research, however, relies on accessibility to jobs metrics that lack
the ability to accurately discern between distinct populations; all jobs are usually counted as being
accessible for everyone, as long as they can be reached.
This study refines the accessibility to jobs concept by developing a measure of accessibility to
low-income jobs for vulnerable residents by public transport. Such a metric has the advantage of
specifically considering the opportunities that can be accessed by vulnerable groups, instead of
grouping together e.g. finance and manufacturing jobs and assuming that these can be filled by
everyone. Furthermore, the metric employs realized public transport travel times by low-income
groups to correctly reflect the different characteristics between mode choice and commutes by
different socio-economic populations. The finer granularity of this metric allows planners and urban
decisionmakers to propose more targeted interventions. The differences between this more
refined metric and the regular accessibility measure are demonstrated by comparing eleven
metropolitan areas across Canada (see figure 1 for an overview of the cities included in this study).
As such, this study is the first to provide a comprehensive look at the state of accessibility in
Canada’s largest cities, and to compare these cities in terms of the equity of their transport and
land use systems.
The rest of this chapter is structured as follows. Section 2 provides a literature study on
accessibility metrics and the relationship between equity and accessibility. In section 3, the data
and methodology used to calculate accessibility to both all jobs and low-income jobs are
explained. Section 4 describes and discusses the results of the accessibility comparisons between
the eleven metropolitan areas, and section 5 concludes this chapter.
6
FIGURE 1 - Overview of the 11 metropolitan areas included in the study
1. LITERATURE REVIEW
Accessibility, or the ease of reaching destinations, is a comprehensive metric measuring the
interaction between land use and transportation (El-Geneidy & Levinson, 2006; Handy & Niemeier,
1997). The concept was first defined by Hansen (1959) (p.73) as “the potential of opportunities for
interaction”, and provides a measure for the variety and number of opportunities, or destinations,
that can be reached from a certain point in space through the transportation network. As such,
accessibility improves upon the concept of mobility by also considering potential and desirable
destinations, instead of only examining an individual’s ability to move.
Accessibility is generally understood to comprise four main components: land use, transportation,
time, and the individual (Geurs & van Wee, 2004). Place-based accessibility metrics measure
accessibility at a certain point in space, and usually only incorporate land use and transport factors
due to data limitations. Person-based metrics, on the other hand, founded in space-time
geography, focus on the individual, and thus also incorporate e.g. time budgets and socio-
economic information (Geurs & van Wee, 2004).
7
Three common metrics of place-based accessibility exist. Cumulative opportunity measures of
accessibility count how many destinations can be reached from a certain point in space within a
certain time threshold using a certain mode, e.g. the number of grocery stores an individual can
reach in 30 minutes by public transportation (Ingram, 1971; Morris et al., 1979; Wickstrom, 1971).
Gravity-based accessibility measures, on the other hand, discount these destinations or
opportunities by distance; the further an opportunity is, the less it contributes to accessibility
(Hansen, 1959; Koenig, 1980). Such measures thus relax the assumption made by cumulative
metrics that individuals stop travelling after a certain time threshold is reached, but are therefore
more difficult to compute and communicate to varying audiences, reducing their chances to impact
policy (Handy & Niemeier, 1997). A third set of commonly used measures are the utility-based
accessibility metrics, which assign each destination with a specific utility and calculate the logsum
of all destinations within a potential choice set (Handy & Niemeier, 1997). Utility-based
accessibility can thus be computed as the denominator of the multinomial logit model. While such
measures require extensive data collection, they can be converted into monetary units using
Hicksian compensation variation, rendering them easier to communicate to urban decisionmakers
(Geurs & van Wee, 2004; Niemeier, 1997).
Accessibility measures were developed with the intention of evaluating transport and land use
systems, while simultaneously allowing for the ability to measure the effects of proposed plans
and investment strategies (Morris et al., 1979). As such, accessibility is designed to accurately
measure the benefits resulting from interacting land use and transport systems. Such benefits
range from increased development potential in highly accessible locations (Ozbay et al., 2003), to
land value premia generated by accessibility (Martínez & Viegas, 2009), decreased risks of social
exclusion (Lucas, 2012), shorter unemployment duration for individuals experiencing high
accessibility levels (Andersson et al., 2014; Korsu & Wenglenski, 2010), and lower unemployment
rates among low-income households (Hu, 2017). Furthermore, accessibility has been linked with
shorter commutes (Levinson, 1998) and, in areas with high accessibility via public transport, higher
incidences of public transport use (A. Owen & Levinson, 2015b).
Accessibility is therefore often used to differentiate the effects of new transport plans between
socio-economic groups: who stands to benefit, and who will lose (Levinson, 2014; Manaugh & El-
Geneidy, 2012)? Driven by such notions of equity, transport researchers have developed a vast
amount of literature discussing accessibility and equity (see, among others, Bocarejo and Oviedo
(2012); Delbosc and Currie (2011); Delmelle and Casas (2012); El-Geneidy, Levinson, et al.
(2016); Foth et al. (2013); Guzman et al. (2017); Lucas et al. (2016); van Wee and Geurs (2011)).
8
Two conflicting notions of equity exist, however. To achieve horizontal equity, accessibility should
be equally divided across the entire population, regardless of individuals’ socio-economic status.
Vertical equity, on the other hand, would be achieved when those with the highest need
experience the highest accessibility; the concept thus posits that socio-economically vulnerable
populations should have higher accessibility (Karner & Niemeier, 2013).
While most scholars focus on measuring accessibility to jobs, accessibility has also been
calculated to health care facilities, schools, grocery stores, and a myriad of other opportunities
(Bissonnette et al., 2012; Grengs, 2015; Guagliardo, 2004; Luo & Wang, 2003; Mao & Nekorchuk,
2013). However, jobs remain the most prominent non-home destinations, and are thus particularly
useful in measuring an area’s attractiveness (A. Owen et al., 2016). Accessibility to jobs, besides
solely giving an indication of how many jobs an individual might reach, also has the added benefit
of providing a measure of nearby amenities; all services we’d like to reach, be they restaurants or
the theatre, for example, employ a certain amount of people and are therefore measured within
the accessibility to jobs framework. While some studies have examined and compared
accessibility to jobs across different cities (A. Owen et al., 2016; Ramsey & Bell, 2014; Tomer et
al., 2011), no such research has been undertaken in the Canadian context.
Despite the prominence of equity in accessibility research, there has been limited focus on
specifically measuring accessibility to low-income or low-wage jobs. Such accessibility metrics
provide a better representation of the state of access for vulnerable groups, as there are often
non-spatial barriers to acquiring high-wage employment (Legrain et al., 2016). While most studies
measuring accessibility to jobs often find that vulnerable groups experience higher accessibility
levels (see for example Foth et al. (2013)), these studies usually do not specifically measure
access to low-income jobs, i.e. to destinations that are more valuable for socially vulnerable
populations (Legrain et al., 2016).
This study contributes to the growing body of literature on accessibility to jobs in three major ways.
Firstly, we move away from the use of arbitrary time limits usually used in cumulative measures
of accessibility, and instead compute the average commute (once for all commuters, and once for
low-income commuters) to determine the time threshold, thereby more accurately modelling
people’s activity spheres. Secondly, along with calculating accessibility to all jobs, we also discern
accessibility to low-income jobs, and specifically measure the latter for vulnerable groups. Finally,
this paper is the first to compare such measures and the consequent equity of the transport and
land use interaction across eleven major Canadian metropolitan areas, in order to provide a more
holistic view on the equity of public transport provision in Canada.
9
2. DATA AND METHODOLOGY
To compute the accessibility to jobs metric, public transportation schedules across the eleven
cities were gathered in the General Transit Feed Specification (GTFS) format from their respective
transport agencies. Transit schedules were obtained for March 2017, or, when not available, for
the available dates closest to this date. For large metropolitan areas with multiple providers, the
schedules from all agencies were obtained with overlapping schedule dates.
Transit schedules were subsequently imported into a geographic information system through the
Add GTFS to a network dataset add-on for ArcGIS. A joint network between the public transport
network and the streets was then created. Travel times between census tract centroids were
computed based on the joint network, through a fastest route calculation during the morning peak
at 8 AM on a regular Tuesday. By creating a joint network, the algorithm would not force the fastest
path to solely choose public transportation: when a route between two census tracts was faster
by walking, the walking route was thus designated as the fastest. The computation for public
transport travel time incorporated walking time, waiting time, and in-vehicle time based on transit
schedules.
Jobs data was acquired through Statistics Canada, in the form of commute tables for each
Canadian province. The tables present the number of commuters, by personal total income and
mode of transport, working in each census tract; both commuters within the same census tract as
those outside are counted. Abstracting unfilled positions, the total amount of individuals working
in each census tract is an accurate proxy for the number of jobs in each tract.
To define the threshold for low-wage jobs, the incomes from the commuting tables were used.
While these might not necessarily perfectly reflect the wage obtained from a certain job (e.g. it
would overestimate a job’s wage for individuals with 2 or more jobs, as their total income would
be higher than any one of the two wages), the commuting tables provide a comprehensive and
uniform way to define a job’s wage across all Canadian provinces and metropolitan areas. To
reflect local variances in the cost of living and wage distribution, the low-income threshold was
determined individually for each metropolitan area as follows: the 30% lowest paying jobs in each
city were designated low-income. As income was only available in brackets, the closest bracket
was chosen as the threshold (e.g. if 29% of the jobs would fall within or below the $20,000-$30,000
bracket, then $30,000 would be the threshold).
Based on census tract travel times and jobs data, accessibility was computed via a cumulative
accessibility metric as follows:
10
�� = � ���
���� �� ���� = �1 � ��� ≥ ���������� 0 � ��� > ����������
where �� is the accessibility in census tract �, �� is the number of jobs in census tract � , ��� is the
travel time between census tracts � and �, and ���������� is the average commute by public
transportation in the region.
Accessibility to low-income jobs was calculated as follows:
�� � = ∑ �""∑ � "" � � ��
���� �� ���� = �1 � ��� ≥ ���������� 0 � ��� > ����������
where �� � is the accessibility to low-income jobs in census tract �, � � is the number of low-income
jobs in census tract � , ��� is the travel time between census tracts � and �, and ���������� is the
average commute by public transportation in the region for those travelling to low-income jobs. ∑ #$$∑ %&$$ represents the ratio of all jobs to low-income jobs in the metropolitan area (close to 10/3 per
definition), and is used to scale accessibility to low-income jobs so that it is directly comparable to
the accessibility to all jobs measure.
The average commute by public transportation in each metropolitan area was computed as
follows. The average travel time for all public transport commuters was calculated by multiplying
the number of individuals using public transport going from each census tract to each other census
tract by the travel time between these census tracts, resulting in total public transport travel time
in the region. This number was subsequently divided by the total number of public transport
commuters to obtain average travel time in the region. For low-income commuters, a similar
method was employed, but only low-income commuters were taken into account in the
calculations. Thus, for each metropolitan area, the average commute for all workers and the
average commute for low-income workers was computed separately. To calculate accessibility,
the average commutes were rounded up or down to the nearest multiple of 5 minutes.
To discern accessibility levels by socio-economic groups, a vulnerability indicator was estimated.
The indicator is composed of the following variables, based on the index for Canadian cities
developed by Foth et al. (2013):
• Median household income
• Unemployment rate
• The percentage of the population that has immigrated within the last 5 years
11
• The percentage of households that spend more than 30% of their total income on housing
rent
The final vulnerability indicator is given by the sum of the z-scores for the latter three variables,
minus the z-score for the median household income. A high indicator therefore reveals high
vulnerability. A census tract was subsequently designated as vulnerable if it was within the 30%
census tracts with the highest indicator.
3. RESULTS AND DISCUSSION
Table 1 presents summary statistics for the 11 metropolitan areas included in this study. Note that
Toronto-Hamilton refers to the entire Greater Toronto and Hamilton Area and thus includes the
Toronto, Hamilton, and Oshawa census metropolitan areas as defined by Statistics Canada. The
11 included areas represent the 10 largest cities in terms of metropolitan population, plus Halifax
to speak for Atlantic Canada. This subset of Canadian cities includes a wide variety of contexts,
from large megalopolises such as Toronto, to smaller regional centres such as London and
Halifax.
However, the size of a metropolitan area does not seem to be reflected in average commute times
– in most cities this figure hovers between 45 and 60 minutes, with Winnipeg residents
experiencing the shortest commutes. This might be explained by better public transport provision
(such as the presence of commuter rail, subways and elevated rail) in larger cities such as Toronto,
Montreal and Vancouver.
On average, it appears that low-income workers have shorter commutes. This difference is
especially profound in Edmonton, where low-income workers travel 10 minutes less than average,
while in London and Kitchener-Cambridge-Waterloo, low-wage commuters are forced to travel
longer than average, although the difference is not large. Thus, computing accessibility at the
same threshold for both low-income workers and all commuters would generally overestimate the
activity spheres of vulnerable populations and thereby result in overestimates of accessibility
levels. As these estimates could then feed into policy, they might result in biased
recommendations that could negatively affect low-income workers.
Note that, in London and Kitchener-Cambridge-Waterloo, low-income workers travel longer on
average, resulting in a higher threshold for the calculation of their accessibility. This raises
questions about choice: are low-income groups in these two cities indeed more willing to travel
longer, or are they travelling longer because their choices are constrained? Comparing with the 9
other cities in this study, the latter seems more plausible; this might indicate a limitation on the use
12
of realized travel times for the calculation of accessibility. Such an approach therefore requires
planners to be inimically knowledgeable about local circumstances to choose appropriate
accessibility indicators.
Figures 2 and 3 present a comparison between accessibility to all jobs and accessibility to low-
income jobs in Canada’s three largest cities: Toronto, Montreal and Vancouver. Average
accessibility levels appear to be related to a city’s size and its number of jobs. Torontonians
experience the largest accessibility levels on average (around 422,000 jobs in an average
commute), while Vancouver inhabitants can only reach 209,000 jobs in an average commute. The
average accessibility level in Montreal is 365,000 jobs.
Distinct patterns of access can be discerned from the maps: accessibility generally drops off with
increasing distance from the central business district, but along rapid transit lines – commuter rail
and the subway in Toronto and Montreal, or the Skytrain in Vancouver – accessibility levels, to
both low-income jobs and all jobs, remain higher, reflecting the benefits conveyed by fast and
frequent modes of transport. While subways seem to generate a linear pattern of high access,
commuter rails only induce high access at stations, mirroring differences in stop spacing between
the two heavy-rail modes. Local employment centres, on the other hand, have a minor effect on
accessibility levels, as the jobs present in these areas are often dwarfed in number by those
located in the central business district, or because these centres were co-located with rapid transit
stops (such as at the Scarborough Centre Station in the Greater Toronto and Hamilton Area).
Interestingly, vulnerable Montreal residents are more fortunate than those in Toronto in terms of
accessibility to low-income jobs: vulnerable Montrealers can access around 133,000 low-income
jobs (449,000 scaled jobs), while vulnerable Torontonians can only reach 93,000 in an average
commute for low-income residents (312,000 scaled jobs), even though the number of low-income
jobs in Toronto is much higher.
13
TABLE 1 - Summary statistics for the 11 metropolitan areas
Metropolitan area P
opul
atio
n
Tot
al n
umbe
r of
jobs
Pop
ulat
ion
dens
ity
(pop
/km
2 )
Med
ian
hous
ehol
d in
com
e (C
AD
)
Une
mpl
oym
ent r
ate
(%)
Ave
rage
com
mut
e (m
in)
Ave
rage
com
mut
e th
resh
old
(min
)
Ave
rage
low
-inco
me
com
mut
e (m
in)
Ave
rage
low
-inco
me
com
mut
e th
resh
old
(min
)
Low
-inco
me
thre
shol
d (C
AD
)
Calgary 1,392,609 585,860 272.5 99,583 9.3 58.29 60 56.84 55 40,000
Edmonton 1,321,426 551,140 140.0 94,447 8.5 63.95 65 53.62 55 40,000
Halifax 403,390 180,860 73.4 69,522 7.3 60.14 60 54.15 55 30,000
Kitchener - Cambridge - Waterloo
523,894 227,500 480.1 77,229 6.4 45.88 45 45.90 46 30,000
London 494,069 199,090 185.6 64,743 7.3 49.23 50 51.42 50 30,000
Montreal 4,098,927 1,750,605 890.2 61,790 7.5 50.51 50 45.85 45 30,000
Ottawa - Gatineau 1,323,783 594,745 195.6 82,053 7.0 54.93 55 49.99 50 40,000
Quebec 800,296 374,680 234.8 65,359 4.6 50.26 50 46.37 45 30,000
Toronto - Hamilton† 7,055,433 2,935,930 862.4 78,437 7.6 58.37 60 52.41 50 30,000
Vancouver 2,463,431 1,004,375 854.6 72,662 5.8 49.86 50 47.41 45 30,000
Winnipeg 778,489 343,365 147 70,795 6.3 43.00 45 42.38 40 30,000 † Refers to the Greater Toronto and Hamilton Area, and includes the Toronto, Hamilton, and Oshawa census metropolitan areas
14
FIGURE 2 - Accessibility to all jobs in Toronto, Montreal and Vancouver
15
FIGURE 3 - Accessibility to low-income jobs in Toronto, Montreal and Vancouver
16
FIGURE 4 - Comparison of accessibility to all jobs and accessibility to low-income jobs
In figure 4, a comparison between average accessibility levels, to all jobs and to low-income jobs,
for all cities included in this study can be seen, which again confirms the difference in access
between Toronto and Montreal as noted above. The blue dots represent the commonly used
accessibility to all jobs metric, while the orange dots show the average of the former metric in
socially vulnerable neighbourhoods. The red dots finally represent the new metric of accessibility
to low-income jobs, averaged for socially vulnerable neighbourhoods.
A city’s size does not fully predict average accessibility levels; in Vancouver, for example, an
average resident can only access as much opportunities as in Edmonton, Ottawa, or Calgary,
which are home to around 700,000 less people and 400,000 less jobs. Similarly, inhabitants of
Kitchener-Cambridge-Waterloo and London have lower accessibility than those in Halifax, even
though the latter city houses 100,000 less residents. The explanation for such discrepancies are
to be found in the particular allocation of land uses and the speed, frequency and coverage of the
transport systems present in these cities.
Note the large differences between the accessibility to all jobs for vulnerable groups (orange) and
accessibility to low-income jobs for vulnerable groups (red). These discrepancies can be explained
by the difference in spatial allocation and distribution of low-income jobs versus all jobs, and by
17
the different activity spheres and thus travel times realized by vulnerable groups. In all cities
(except for Kitchener – Cambridge – Waterloo), accessibility for vulnerable individuals would have
been overestimated with the commonly used all jobs metric. In Toronto, for example, policy
makers only having access to an all jobs metric would conclude that vulnerable groups are better
off than other residents. However, focusing only on low-income jobs that can be accessed by
these individuals, it is clear that this is not the case.
In all cities, except for Toronto, Calgary, Edmonton, and Quebec, vulnerable groups still have
access to a larger number of relevant job opportunities than average; vertical equity thus appears
high in most Canadian metropolitan areas. Nevertheless, the variation between accessibility to all
jobs and accessibility to low-income jobs for vulnerable groups (especially in Toronto and
Montreal) highlights the benefits of calculating these two metrics separately.
To further illustrate the importance of distinguishing between the two accessibility metrics, we
plotted accessibility against transit modal share, for both the regular accessibility measure, and
the measure of accessibility to low-income jobs for vulnerable residents, see figure 5. Transit
accessibility has been shown to be correlated with public transport mode share (A. Owen &
Levinson, 2015b), and as such the effects of access on mode share provide a representative
example demonstrating the significance of employing two separate accessibility metrics. The
difference between the two accessibility levels can again be distinguished – accessibility to low-
income jobs for vulnerable groups is, for most cities, higher than the regular accessibility metric.
Importantly, the effect of the accessibility metric on transit modal share differs between the two
measures. While for every increase of 100,000 jobs in the regular accessibility measure, modal
share increases by 4.25%, the comparable figure for the low-income accessibility metric for
vulnerable groups is 6.21%. Recall that low-income accessibility was scaled by the ratio of all jobs
to low-income jobs, and is therefore directly comparable to accessibility to all jobs; the contrast in
effect sizes is therefore not due to a lower number of low-income jobs. Thus, vulnerable
populations appear to be more sensitive to changes in accessibility to jobs they can access – had
decisions been based on the regular accessibility metric, they would have considerably
underestimated the effect of increasing accessibility to low-income jobs for vulnerable residents.
Future research should examine the effects of accessibility to low-income jobs in further detail.
18
4. CONCLUSION
Accurate policy recommendations require the creation of adequate and robust performance
indicators that can precisely measure progress. This study has therefore developed the metric of
accessibility to low-income jobs for vulnerable residents, which, in contrast with commonly used
measures of accessibility to all jobs, provides policy makers and urban planners with a more fine-
grained tool to examine the effects of new transportation plans and projects across different socio-
economic populations. In effect, this detailed accessibility measure provides a first step towards
the segmentation of different groups within accessibility literature and practice – a common
occurrence in studies on the different types of cyclists and public transport users (Damant-Sirois
et al., 2014; Jensen, 1999; Krizek & El-Geneidy, 2007). Through such a segmentation approach,
the benefits of both place-based and people-based accessibility metrics can be reconciled, namely
the communicability and data requirements of place-based measures, combined with the detail
common in people-based metrics.
This study has compared accessibility to all jobs by public transport, and the new metric of
accessibility to low-income jobs for vulnerable groups for 11 metropolitan areas in Canada,
ranging from large metropolises to smaller regional cities. On average, Toronto residents
experience the highest accessibility levels. When focusing on low-income jobs, however,
vulnerable groups are far better off in Montreal in terms of accessibility, even though the total
number of low-income jobs is much lower than in Toronto. Overall, it appears that vulnerable
individuals experience higher accessibility levels than their fellow residents in their respective
FIGURE 5 - Accessibility and public transport mode share
19
cities, although this trend does not appear in Toronto, Calgary, Edmonton, and Quebec; these
latter cities thus seem to lag behind their counterparts in terms of vertical equity. The
discrepancies between the two metrics were further highlighted by comparing the effects of access
on public transport mode share: it is clear that vulnerable residents are more sensitive to
accessibility changes. Planners and urban decisionmakers, and in turn their policy
recommendations, thus stand to benefit greatly from employing more detailed and segmented
accessibility metrics.
20
CHAPTER 2: UNDERSTANDING THE RELATIONSHIP BETWEEN CHANGES IN ACCESSIBILITY TO JOBS, INCOME AND
UNEMPLOYMENT IN TORONTO, CANADA2
In many urban areas, transport agencies are trying to provide all citizens with greater access to
opportunities as a means to improve residents’ well-being (Boisjoly & El-Geneidy, 2017b; Handy,
2008; Proffitt et al., 2015). Several cities particularly intend to increase access to opportunities in
socially deprived areas, in order to support social inclusion and enhance the quality of life of
residents in these neighbourhoods (NSW Government, 2012; San Diego Association of
Governments, 2011; Transport for London, 2006). In this context, research suggests that
improvements in access to opportunities by public transport can bring considerable benefits to
vulnerable populations, as they are more likely to rely on this mode for accessing their destinations
(Stanley & Lucas, 2008).
To quantify access to opportunities, accessibility, or the ease of reaching destinations, is
increasingly being used in research and practice as a key land use and transportation performance
measure. From a social equity perspective, accessibility has been used as a tool to assess the
socio-spatial distribution of public transport services (Bocarejo & Oviedo, 2012; Delmelle & Casas,
2012; Golub & Martens, 2014; Kawabata & Shen, 2007), and to evaluate how changes in
accessibility differ across socio-economic groups as a result of projected or new infrastructure
projects (Foth et al., 2013; Manaugh & El-Geneidy, 2012; North Central Texas Council of
Governments, 2016; Paez et al., 2010; Southern California Association of Governments, 2016).
While a large body of literature has assessed accessibility levels for different socio-economic
2 Co-authored with Geneviève Boisjoly and Ahmed El-Geneidy. Paper presented at the 2018 Annual TRB Meeting, Washington DC.
21
groups, or changes in these accessibility levels over time, little research has been conducted to
assess the outcomes of such improvements in accessibility.
The goal of this study is, therefore, to assess the relationship between improvements in the levels
of accessibility to jobs by public transport and the resulting socio-economic benefits, measured by
changes in median household income and unemployment rate over time in the Greater Toronto
and Hamilton Area, Canada. For this purpose, competitive accessibility levels to employment
opportunities by transit and by car are calculated for all census tracts in 2001 and 2011. The
vertical equity of accessibility by transit is then assessed for both years by comparing accessibility
levels across median household income deciles. Two linear regressions are subsequently
performed to examine the relationship between accessibility changes and income and
unemployment at the census tract level, while controlling for the movement of residents. This study
contributes to the literature on accessibility and the equity of outcome resulting from these
accessibility levels, and is of relevance to planning professionals and researchers wishing to
investigate the effects of accessibility improvements across neighbourhoods, especially low-
income ones.
The rest of the chapter is organised as follows. Section 2 explains the concept of accessibility,
examines how equity is incorporated in academic literature on this concept, and presents previous
literature on accessibility, employment and income. Section 3 considers the data and methodology
used to investigate the relationship between improvements in transit accessibility and changes in
income and unemployment, and section 4 presents and discusses the findings. Section 5 then
concludes the chapter and provides recommendations for further research.
1. EQUITY OF ACCESSIBILITY AND EQUITY OF OUTCOME
1.1 Accessibility
Accessibility was first defined by Hansen (1959) (p.73) as “the potential of opportunities for
interaction”. In contrast with mobility, accessibility also considers land use factors such as the
variety and number of destinations that can be reached, instead of only examining an individual's
ability to move through the transportation network (Handy & Niemeier, 1997). Geurs and van Wee
(2004) posit that accessibility measures should comprise four interacting components: land use,
transportation, time, and the individual. Accessibility thus tries to incorporate the spatial distribution
of activities, the transport system connecting these activities, the time constraints of individuals
and services, and personal needs and abilities to provide a more accurate picture of the
performance of transport systems.
22
There are several commonly used measures of accessibility, most of which take into account only
the land use and transportation component, as they can be more easily computed, interpreted,
and communicated, increasing their chances to impact policy (Geurs & van Wee, 2004; Handy &
Niemeier, 1997). Cumulative measures of accessibility count the number of opportunities that can
be reached within a set time-frame, for example the number of jobs an individual can reach within
45 minutes of travel (Wickstrom, 1971). Gravity-based accessibility measures, on the other hand,
take into account that people will not stop travelling at an arbitrary time-limit, and weigh
opportunities by distance; the further an opportunity is, the less it contributes to accessibility
(Hansen, 1959). While more realistic, gravity-based measures require the prediction of a distance
decay function, rendering them more difficult to communicate, interpret and analyze across
studies.
To account for competition effects, for example among workers competing for jobs, the concept
of accessibility has also been extended to include measures of competitive accessibility (Q. Shen,
1998). As cumulative and gravity-based accessibility only measure the ‘supply side’ of
opportunities (Geurs & van Wee, 2004; Morris et al., 1979), they assume that no capacity
limitations exist. Therefore, when accessibility to jobs is examined through the lens of ordinary
cumulative or gravity-based accessibility measures, it is assumed that one job can be filled by an
infinite number of workers. To more accurately reflect reality, a demand potential is first computed
by determining how many individuals can access each opportunity. Each opportunity is then
discounted by this demand potential when calculating accessibility using the cumulative or gravity-
based approach in what is known as a competitive measure of accessibility (Q. Shen, 1998).
1.2 Equity of accessibility
Measures of accessibility have often been used to consider the equity of the joint benefits provided
by the land use and transportation system (see for example (Delmelle & Casas, 2012; Golub &
Martens, 2014; Grengs, 2015; Guzman et al., 2017)). Two different interpretations of equity in
accessibility research exist, both founded in the ethical concept of egalitarianism (Foth et al., 2013;
van Wee & Geurs, 2011). Horizontal equity requires that all members of society have equal access
to all resources. Vertical equity, on the other hand, implies that the more vulnerable groups should
be granted more resources. From this point of view, it would be more beneficial to society to
increase the accessibility of unemployed young individuals than to increase the accessibility of
wealthier individuals (Lucas et al., 2016). Yet another approach defines an equitable system as
having a minimal gap between transit and car accessibility (Golub & Martens, 2014; Karner &
23
Niemeier, 2013), after which both the horizontal and vertical equity of the distribution of this gap
can be measured.
Current literature mostly focuses on examining the vertical equity impacts of transportation
projects. To examine this type of equity, socially vulnerable groups first need to be defined.
Several studies identify socio-economic groups based solely on income (for example (Fan et al.,
2012; Guzman et al., 2017)), whereas other studies also examine race, poverty status, minorities,
and housing characteristics (Delmelle & Casas, 2012; Golub & Martens, 2014; Grengs, 2015), or
create a social indicator combining several of these measures (Foth et al., 2013). The vertical
equity of accessibility can then be investigated by comparing accessibility levels across different
populations.
A distinction is often made between equity of opportunity and equity of outcome (Delbosc & Currie,
2011; Litman, 2002; van Wee & Geurs, 2011). Studies discussing the horizontal and vertical equity
of accessibility address equity of opportunity, but refrain from making judgements on the outcome
of the process. This paper attempts to connect the two concepts by considering the link between
equity of opportunity, measured by accessibility, and equity of outcome, measured by changes in
unemployment and income over time.
1.3 Accessibility, unemployment and income
To determine the outcomes and subsequent benefits resulting from accessibility and accessibility
changes, previous studies have focused on examining the relationship between accessibility to
jobs and socio-economic status, mostly concentrating on unemployment duration. Korsu and
Wenglenski (2010), using micro-data, demonstrate that low accessibility to jobs is related to high
unemployment in Paris, and find that workers living in areas with very low accessibility have a
1.7% higher probability of being unemployed for longer than one year compared to workers living
in neighbourhoods with medium accessibility. To this end, the authors use a measure of
cumulative accessibility, by public transport or car depending on car ownership, specifically
considering the employment opportunities of the same socio-professional status as the individuals
in question. Andersson et al. (2014) investigate low-income workers who were subject to mass
layoffs in several US cities, and find that high accessibility to jobs is associated with a reduction
in the time spent looking for work. A competitive measure of accessibility to low-income jobs is
used for this purpose, taking into account the probability of using car or public transport, and
explicitly considering competing job searchers to account for labour market tightness. Tyndall
(2015) notes that after the closure of the R train in Brooklyn due to hurricane Sandy,
24
unemployment rates along the line increased considerably, especially for those without a private
vehicle, demonstrating that substantial changes in the public transport system affect
unemployment. This study did not, however, examine the accessibility impacts of this endogenous
shock to the transport system. Blumenberg and Pierce (2014) find that living close to a bus stop
highly increases the chances of maintaining consistent employment, while having access to a
private automobile has also been shown to be related to increased employment (Blumenberg &
Pierce, 2017). Larson (2017) examines the relationship between access to jobs by public transport
(broadly defined as the observed transit modal share) and economic opportunity over four
decades in four US cities, and concludes that there is a positive relation between transit access
and economic opportunity in predominantly white neighbourhoods in Orlando and Minneapolis,
while a similar relationship is present in non-white areas in Birmingham.
This emerging body of literature suggests that accessibility to jobs is a potential determinant of
unemployment duration. However, little is known about the relationship between unemployment
rates and accessibility over time at a more aggregate, metropolitan scale; the literature presented
above has not examined how accessibility changes impact longer term unemployment duration
and more aggregated unemployment rates. Furthermore, no study has, to our knowledge,
examined changes in accessibility and median household income over time. To provide a more
holistic view on the relationship between accessibility changes and consequent changes in socio-
economic status at an aggregate level, this study attempts to investigate the change in both the
unemployment rate and median household income over a ten-year period. This paper therefore
contributes to the literature by presenting a long-term study associating a robust accessibility
measure with equity of outcome.
2. DATA AND METHODOLOGY
2.1 Study context
The Greater Toronto and Hamilton Area, the most populous metropolitan region in Canada,
housing 5.6 million residents in 2001 and 6.6 million inhabitants in 2011, was chosen to examine
the relationship between transit accessibility improvements and changes in income and
unemployment. The region is well connected by public transport, and is home to a subway,
commuter train system and bus network (Figure 6). While the subway only serves the City of
Toronto, the bus and train network extend across the entire region. During the ten-year study
period, several infrastructure projects altered the public transport network in the area. In 2002, a
new subway line, the Sheppard line (the line shown in green in figure 6), was opened, serving five
25
new stations in the north of the City of Toronto. Additionally, several new train stations were
constructed and new express bus services were introduced. At the same time, transit mode share
increased from 20% in 2001 to 21% in 2011.
2.2 Data
Three different data sources were used for the analysis. Census and employment data for 2001
and 2011 were obtained from Statistics Canada. This data was enriched by a cumulative
accessibility measure for a 45-minute trip by transit in 2011 at the census tract level, derived from
GTFS data. The third data source, Metrolinx, provided travel time from 2001 at the traffic analysis
zone (TAZ) level, calculated through the EMME travel demand modelling software, for both public
transportation and automobile. Additionally, car travel time from 2011 during the AM peak was
also supplied by Metrolinx.
A competitive measure of accessibility for 2001 at the TAZ level was first calculated using 2001
travel times and employment. Competitive accessibility is given by:
�'� = ∑ #()(�+(,).(,� , where /�' = ∑ �0�� (���')
�'� reflects the accessibility at point i for transportation mode m, �� is the number of opportunities
at location j, and (���') is 1 when the travel time between locations i and j (���') is smaller than the
set-time limit, and 0 otherwise. /�' represents the demand for the opportunities at location j, and
is given by the total labour force (�0�) that can access those opportunities within the set time-limit.
To ensure consistency with available data from 2011, and to allow for comparisons, the
accessibility measure was calculated for a 45-minute trip limit for public transport, and a 30-minute
limit for car, and then projected into 2011 census tract boundaries through a nearest neighbour
interpolation. These time limits reflect the average commute times in Toronto for both modes (49
and 29 minutes respectively (Statistics Canada, 2010)), in order to capture the opportunities an
individual can access in an average trip, while accounting for competition from other residents
trying to reach the same opportunities.
26
FIGURE 6 - Context map
2.3 Methodology
To investigate the relationship between improvements in transit accessibility and changes in the
unemployment rate and median household income, two linear regression models are employed.
The first model predicts median household income in 2011, based on median household income
in 2001 and changes in accessibility by car and transit between the two years. The second model
is specified in a similar manner: the unemployment rate in 2011 is related to the unemployment
rate in 2001 and changes in accessibility levels.
As changes in income, especially for low income census tracts, could be related to gentrification,
i.e., the upgrading of the socio-economic status of a neighbourhood through local migration
(Lyons, 1996), several additional variables are added to the model. Literature on the relation
between transit and gentrification usually investigates land and housing values, changes in
income, race, car ownership, the number of professionals, and educational attainment to identify
27
gentrifying areas (Grube-Cavers & Patterson, 2015; Kahn, 2007; Pollack et al., 2011). A
neighbourhood is said to be gentrifying if these variables change faster than the average in the
metropolitan area. Such an approach, however, does not account for the movement of people.
Some of the changes noted by the literature could, instead of being linked to gentrification, have
resulted from an improvement in the conditions of the individuals living in a certain neighbourhood,
without the presence of outside forces pushing these residents out; increases in income do not
always imply that people were pushed out and wealthier individuals moved in (Freeman, 2005).
Also incorporating the percentage of people moving mitigates these disadvantages and
acknowledges that in-movers are the driving force behind gentrification (Freeman, 2005).
Consequently, the change in the percentage of residents with a bachelor’s degree or higher, and
the percentage of residents that have moved between 2006 and 2011 are included in the
regression model to control for the effects of gentrification, and, more broadly, migration. The
summary statistics of the variables used in the two models are shown in table 2.
TABLE 2 - Summary statistics
Variable Mean Standard dev.
Median Household Income in 2011 ($1,000) 75.664 26.536 Median Household Income in 2001 ($1,000) 64.534 21.558 Unemployment rate in 2011 (%) 8.7173 3.1598 Unemployment rate in 2001 (%) 5.7868 2.4814 Change in competitive accessibility by transit (jobs/worker) -0.0897 1.1893 Change in competitive accessibility by car (jobs/worker) 0.2422 0.2917 Change in percentage of residents with a bachelor’s degree or higher (%)
4.3710 4.9699
Percentage of residents that have moved between 2006 and 2011 (%)
35.131 11.480
3. RESULTS AND DISCUSSION
Figure 7 shows the spatial distribution of median household income and the unemployment rate
in the GTHA in 2001 and 2011. In the top two maps, the lightest colour represents the census
tracts with the lowest income, whereas the darkest color represents the least vulnerable
neighbourhoods. In both years, the low-income census tracts are centred in a ring around
downtown Toronto, although a suburbanization of low income areas has occurred; the
neighbourhoods to the north and east of the City of Toronto have become more vulnerable in
2011. The outer suburbs, as well as the CBD of Toronto, house higher income populations in both
years. In the bottom map, the lowest unemployment rate is presented in the lightest color, while
the highest unemployment rate is shown in the darkest color. The financial crisis of 2007-2008
28
radically changed the pattern of unemployment across the region: the unemployment rate
skyrocketed between 2001 and 2011 in almost every census tract, especially in the outer suburbs.
The spatial distribution of competitive accessibility by public transport and car in both 2001 and
2011 are shown in figure 8. Transit accessibility was calculated for a maximum travel time of 45
minutes, whereas car accessibility was computed for a 30-minute trip. The two modes display
profoundly different spatial patterns, due to significant directionality present in the public transport
system. During the morning peak, the GO train network focuses on bringing residents into the
Toronto CBD, while the service in the opposite direction is close to non-existent. Suburban job
centres are therefore protected from competition by transit: only local residents can access these
employment opportunities, resulting in high competitive accessibility levels. Competitive
accessibility by transit is thus mainly determined by competition effects. In contrast, accessibility
by car is mostly influenced by the presence of job opportunities, as directionality is less present in
the highway and street networks. Car accessibility is thus highest in downtown Toronto, where the
largest amount of job opportunities is present. Between 2001 and 2011, accessibility by private
automobile rose substantially in Toronto and in the western parts of the region, whereas a small
decrease was observed in the eastern census tracts. At the same time, competitive accessibility
by transit increased in a few clusters of suburban job centres, and decreased in the rest of the
Greater Toronto and Hamilton Area.
29
FIGURE 7 - Median household income and unemployment rate in the GTHA in 2001 and 2011
30
FIGURE 8 - Transit accessibility in the GTHA in 2001 and 2011
31
3.1 Vertical equity
Figure 9 presents transit accessibility standardized values (z-scores) by income decile. In 2001,
the four deciles with the lowest income in the region experience considerably higher competitive
accessibility levels by transit than all other groups, highlighting that accessibility is vertically
equitable in the GTHA, which is consistent with the findings of Foth et al. (2013) for the Greater
Toronto and Hamilton Area. Competitive accessibility of the four groups with the lowest income
decreased between the two years, however, although they continue to have a considerably higher
accessibility than the other income deciles. The investments in commuter trains, connecting
wealthier neighbourhoods to downtown Toronto, have therefore succeeded in increasing
accessibility to employment for high income census tracts. This suggests that, while the vertical
equity of the transportation and land use system is still high in the GTHA, there is a trend towards
decreasing vertical equity and increasing horizontal equity. Note that, as socially vulnerable
groups have lower car ownership (Potoglou & Kanaroglou, 2008), this decrease in accessibility
can result in substantial negative consequences for the region’s most vulnerable populations. To
quantify the effects of these accessibility changes on neighbourhood socio-economic status,
results of the linear regression models are presented in the next section.
FIGURE 9 - Relative competitive accessibility by transit, by income decile in the GTHA
32
3.2 Linear regression models
Table 3 shows the results of the two linear regression models, with both models showing similar
patterns. Only the variables that are statistically significant will be described here. The model
predicting median household income in 2011 demonstrates that higher median household income
in 2001 is associated with higher median household income in 2011, while the coefficient of 1.12
for this variable suggests that overall income levels rose by 12% during the study period, while
controlling for all other variables present in the model. Changes in competitive accessibility by
transit, and the interaction term between this variable and median household income in 2001, are
significantly related to income in 2011. For example, a census tract with a median household
income of $40,000 in 2001 is predicted to have an extra increase in income of (7.67 – 0.099*40)
= 3.71 ($3,710) in 2011 per extra unit in competitive accessibility (Table 3). A one unit increase in
competitive accessibility occurs when a person can access an extra job that is not accessible to
all other residents in the region. The effect of competitive accessibility reverses when income in
2001 is higher than $77,475. As higher income populations are more likely to move to less dense
areas in search for open space, they tend to migrate to areas without public transport access. As
a result, median income decreases in areas where these wealthy groups move out. Increases in
competitive accessibility by car are also statistically significant and associated with higher incomes
in 2011: a one unit increase in car accessibility is predicted to increase income by $3,370. An
interaction term between car accessibility and baseline household income in 2001 was also
analyzed, but was not significant, indicating that the effect of accessibility by car is income-
independent.
The remaining statistically significant coefficients highlight that increases in the percentage of
residents with a bachelor’s degree or higher, and stable neighbourhoods (without many people
moving) are related to higher median household incomes in 2011. The coefficients for accessibility
changes by both car and public transport highlight that changing equity of opportunity, measured
by accessibility, is associated with a changing equity of outcome, measured by income.
The second model indicates that higher unemployment rates in 2001 are associated with higher
unemployment rates in 2011, suggesting that census tracts with high unemployment rates in 2001
still have higher unemployment in 2011. An extra accessible job by transit that cannot be reached
by any other individual (a one unit increase in transit accessibility) is related to a 2.5 percentage
point decrease in unemployment rate for census tracts with a median household income of $0. If
median household income in 2001 increases, the effects of changes in transit accessibility lessen
and reverse at a median household income of $78,052. In contrast, the change in car accessibility
33
has a uniform effect across income: one extra accessible job by car that cannot be reached by
others is linked to a decrease of 0.54 percentage points in unemployment rate. As with the model
predicting income, increases in the percentage of residents with a bachelor’s degree or higher are
significantly associated with lower increases in the unemployment rate. These results are
consistent with the findings presented by Tyndall (2015), who found that a substantial change in
the provision of public transport (and thus a considerable change in access by transit) was
associated with changing unemployment. This suggests that the conclusions by Korsu and
Wenglenski (2010) and Andersson et al. (2014) can be extended from unemployment duration at
the individual level to aggregated unemployment rates at the neighbourhood scale.
Table 4 presents predicted values for median household income and the unemployment rate in
2011 for all income deciles in 2001. The values are predicted for a constant transit accessibility,
and for a transit accessibility that increased by one unit during the study period. Median household
income in 2011 is greater for all deciles except the two wealthiest groups if accessibility by public
transport increased instead of remaining constant. The premium generated by transit accessibility
ranges from $3,812 for the lowest income decile to -$13,744 for the highest income decile. A
similar pattern is present in the predicted unemployment rates: the predicted effect of a unit
increase in competitive accessibility by transit is -1.28 percentage points for the poorest census
tracts, and 4.52 percentage points for the wealthiest decile. Based on these predictions, we can
infer that the decreasing vertical equity of transit accessibility (as shown in figure 9) is associated
with a widening of the income gap in the GTHA.
34
TABLE 3 - Regression results for census tract median household income and unemployment rate in 2011 in the Greater Toronto and Hamilton area
Income Unemployment rate
Variable Coefficient Sig. Confidence
interval†
Coefficient Sig. Confidence
interval†
Constant 5.11 *** 2.071 8.15 4.7788 *** 4.2652 5.2925 Median household income in 2001 1.121 *** 1.093 1.149 - - - - Unemployment rate in 2001 - - - - 0.6986 *** 0.6362 0.761 Change in accessibility by transit 7.67 * 1.276 14.065 -2.5523 ** -4.2517 -0.8529 Change in accessibility by transit • Median household income in 2001
-0.099 * -0.181 -0.016 0.0327 * 0.0108 0.0546
Change in accessibility by car 3.37 *** 1.49 5.249 -0.5402 ** -1.0368 -0.0436 Change in percentage of residents with a bachelor’s degree or higher
0.664 *** 0.554 0.775 -0.093 *** -0.1232 -0.0627
Percentage of residents that have moved between 2006 and 2011
-0.154 *** -0.206 -0.103 0.0116 -0.0020 0.0252
Adjusted R2 0.8695 0.352
Dependent Variables: Median household income in 2011 ($1,000), Unemployment rate in 2011 (%) * 95% significance level | ** 99% significance level | *** 99.9% significance level † 95% confidence interval
35
TABLE 4 - Predicted 2011 income and unemployment rates for each income decile in 2001
Change in transit accessibility = 0 Change in transit accessibility = 1
Income
decile
Income
2001
Unemployment
rate 2001
Predicted
income 2011
Predicted
unemployment rate
2011
Predicted
income 2011
Predicted
unemployment rate
2011
1 38,967 9.7260 47,100 11.4435 50,913 10.1655
2 45,353 7.5418 54,260 9.9177 57,440 8.8484
3 50,835 6.5180 60,404 9.2024 63,042 8.3124
4 57,487 5.8651 67,860 8.7463 69,839 8.0738
5 63,125 5.6117 74,182 8.5693 75,603 8.0812
6 70,204 5.0530 82,117 8.1790 82,837 7.9223
7 75,605 4.6826 88,172 7.9202 88,357 7.8402
8 81,954 4.6638 95,289 7.9071 94,846 8.0347
9 89,749 4.1651 104,026 7.5587 102,811 7.9411
10 216,308 4.0577 245,900 7.4837 232,155 12.0046
4. CONCLUSION
Accessibility to jobs by public transport is a key factor explaining the quality of life of individuals.
Results show that accessibility to jobs by public transport is relatively vertically equitable in the
Greater Toronto and Hamilton Area, although vertical equity decreased between 2001 and 2011.
The census tracts with the lowest income boast the highest accessibility to jobs thanks to their
proximity to downtown Toronto and the public transport network, while wealthier groups
experience lower accessibility levels.
This study suggests that, for low and medium income census tracts, increases in transit
accessibility are related to higher increases in income. For wealthier census tracts, increases in
transit accessibility are associated with decreases in income, potentially due to the migration of
high-income populations to less dense neighbourhoods, away from transit. The change in
accessibility by car, on the other hand, has a uniform effect across income deciles and is
associated with larger income increases. The equity of accessibility to employment opportunities
thus plays a key role in determining resulting equity of outcome, stressing the need for methods
that can incorporate equity considerations into the evaluation of new transportation projects.
It is important to note that the findings from this study are not conclusive, nor can they determine
a causal relationship; more analysis is needed in multiple cities across the globe to further
investigate the relationship between accessibility improvements and changes in income and
unemployment. While multiple variables related to migration were examined, this study does not
36
fully capture the impacts of population movement between 2001 and 2011. The uncovered
relationship could therefore potentially be explained by transit accessibility attracting medium
income populations, resulting in increases in income for low income areas, and decreases in
income for the wealthiest neighbourhoods. This highlights the need for further research in order
to disentangle the complex socio-spatial relationships uncovered in this study. Ideally, future
research should employ micro-data to track individuals over time, and use surveys and interviews
to shed more light on individual changes in accessibility and socio-economic status.
Future studies should also include the cost of transportation in their analysis and normalize the
fares according to income. This would lower the accessibility of the entire population (El-Geneidy,
Levinson, et al., 2016), and could reduce accessibility for socially vulnerable groups compared to
wealthier groups.
Different types of jobs were not distinguished in the present study, although people cannot access
all the different jobs that exist within a city; an individual without a high school diploma will not be
able to access the high-wage service-sector jobs that cities offer, regardless of the transport and
land use system. Future studies should therefore differentiate low, medium, and high income jobs
when comparing accessibility across different groups and different years. The analysis should
also take into account the time when different jobs start and incorporate the time aspect in the
calculation of accessibility by public transport.
Nevertheless, the results of this study demonstrate a clear association between improvements in
accessibility by transit and positive outcomes (measured by changes in income and
unemployment) for neighbourhoods with low and medium income. The relationship uncovered in
this study establishes new directions for future research in order to explore the equity of outcome
resulting from changing accessibility levels.
ACKNOWLEDGEMENTS
This research was funded by the Social Sciences and Humanities Research Council of Canada
(SSHRC) program. We would like to thank David King, Nicholas Day and Joshua Engel-Yan from
Metrolinx for providing the car travel time matrix in 2011. The authors would also like to thank
David Levinson and Emily Grisé for their helpful insights.
37
CHAPTER 3: ACCESSIBILITY-ORIENTED DEVELOPMENT3
Local authorities worldwide continue to pursue transit-oriented development (TOD) strategies in
order to increase transit ridership, curb traffic congestion, and rejuvenate urban neighbourhoods
(Cervero et al., 2002; Curtis et al., 2009; Papa & Bertolini, 2015; Ratner & Goetz, 2013). For years,
TOD has garnered attention by scholars and transport professionals alike (Calthorpe, 1993; City
of Denver, 2014; Gilat & Sussman, 2003). Neighbourhoods are often defined as TODs when they
are situated close to transit, allow for higher density development, and possess diversified land
uses (Cervero et al., 2004; Kamruzzaman et al., 2015). TOD therefore not only involves the
construction of public transport infrastructure and provision of service, but also requires the
integration of transport and land use (Bertolini et al., 2012; Jacobson & Forsyth, 2008); in this way,
TOD intends to achieve a holistic way of compact urban development, enabled by supporting
public sector policies such as zoning and tax incentives. As TODs usually also encompass
increased attention to urban design, livable spaces and walkability, the demand for housing in
TOD areas results in increased premiums for homes located in TODs (Duncan, 2011; Mathur &
Ferrell, 2013; Renne, 2009). Residents in these areas have also been found to rely more on transit
and active modes of transport, seemingly fulfilling the promises of TOD (Chatman, 2006;
Kamruzzaman et al., 2015), although the relationship between TOD and transit use has been
found to differ between trip motives (Langlois et al., 2015), and not the ‘T’ in TOD, but rather limited
parking availability and higher density may be causing the observed decrease in car use
(Chatman, 2013).
Areas planned as TOD do not always function as foreseen; in many cities, development on
planned sites has been close to non-existent. One potential reason is that the connection between
the (planned) transit investment and land use at both the local and broader spatial scale are often
3 Co-authored with Ahmed-El Geneidy and David Levinson. Paper presented at the 2018 Annual TRB Meeting, Washington DC. This chapter is currently under review as a manuscript at
Transport Geography
38
overlooked. At the local scale, transit-adjacent developments (TADs) may fail to take advantage
of their proximity to transit and bring almost none of the benefits normally associated with TODs
(Renne, 2009). The often physical nature of the definition of TODs (‘density near transit’)
contributes to this problem (Belzer & Autler, 2002).
However, it is at the regional scale that the TOD concept tends to break down more often. TOD is
an inherently local planning tool, and does, at its core, not consider regional land use patterns.
While regional approaches to TOD planning have been proposed (see e.g. P. Newman (2009);
Staricco and Vitale Brovarone (2018)), they are not sufficient to combat this issue and their use is
far from widespread. The TOD concept, even in its regional variant, only considers access to
transit, but not the accessibility that is provided by transit (i.e., what destinations does the transit
service allow me to access?) (Belzer & Autler, 2002; Guthrie & Fan, 2016; Renne, 2009). As travel
patterns are mostly determined by the region-wide levels of accessibility provided by transport
systems, the use of TOD, as such, is insufficient to increase transit usage (Boarnet, 2011;
Chatman, 2013) and to attract urban development. We contend that these issues can be alleviated
by introducing the concept of accessibility-oriented development (AOD).
AOD will help planners to explicitly consider not just access to transit, but also the accessibility
provided by transit. Accessibility, or the ease of reaching destinations, is an easy-to-use measure
that can help unravel the intricacies involved in combined land use and transport planning in the
minds of planning professionals and urban decision makers (Boisjoly & El-Geneidy, 2017a).
Access to destinations is usually operationalized as the number of destinations that can be
reached from a certain point in space. As such, accessibility recognizes the inherent connection
between transport and regional land uses (in the form of destinations) and can be used to
overcome the local focus of TODs.
We define accessibility-oriented development as a strategy that balances accessibility between
employment opportunities and workers to foster an environment conducive to urban development.
AOD occurs both naturally through the market and with direction from planners. The AOD concept
invites planners to leverage access to steer, slow down, or speed up the phenomena that naturally
follow from accessibility changes, namely changing commute times and economic development.
AOD areas are therefore neighbourhoods or sites where planners are using the various tools at
their disposal to control accessibility levels in order to attract a particular mix of residential,
commercial or industrial development. We hypothesize that transport investments made on the
principles of AOD planning will naturally result in development occurring in the targeted
39
neighbourhoods, and, through lower commute times, a better quality of life for residents. This
study aims to test the hypotheses underlying the accessibility-oriented development concept.
The rest of this chapter is organised as follows. Section 2 describes the concept of accessibility
and links it with economic development. Section 3 defines AOD more thoroughly and assesses
the validity of using AOD, by testing the two underlying hypotheses in a case study of the Greater
Toronto and Hamilton Area, Canada, using access to jobs and workers in 2001 and 2011. In
Section 4 the results of the regression models testing AOD are discussed. Section 5 then
concludes the chapter and provides policy recommendations for the implementation of AOD.
1. LITERATURE
1.1 Accessibility
Accessibility is a comprehensive measure of the land use and transport interaction in a region and
illustrates the ease of reaching destinations (Geurs & van Wee, 2004; Handy & Niemeier, 1997).
Accessibility was first defined by Hansen (1959), who used the measure to develop a residential
land use model, under the assumption that accessibility was a main driver of residential
development. This paper builds on this seminal work by testing the relationship between
accessibility and development across different modes in the Greater Toronto and Hamilton Area.
Two measures of accessibility are widely employed. Cumulative opportunity measures of
accessibility compute how many opportunities an individual can reach within a predefined time
threshold (Wickstrom, 1971), whereas gravity-based (or, equivalently, time-weighted cumulative
opportunity) accessibility measures relax the assumption that people only travel until an arbitrary
threshold, and discount opportunities by distance (or time) (Hansen, 1959). While gravity-based
measures of accessibility more realistically reflect behavior, they require the prediction of a
distance decay function and are thus more difficult to calculate, communicate, and compare
across studies (El-Geneidy & Levinson, 2006).The concept of accessibility has been widely used
to shed light on the benefits resulting from land use and transport systems. These benefits range
from higher land values (El-Geneidy, van Lierop, et al., 2016), over smaller risks of social exclusion
(Lucas, 2012), to shorter unemployment duration (Andersson et al., 2014; Korsu & Wenglenski,
2010) and increased odds of firm birth in areas with high accessibility levels (Holl, 2004).
Furthermore, access by public transport has been shown to be related to increased transit mode
share (A. Owen & Levinson, 2015a). Accordingly, to measure how these benefits are distributed
across different socio-economic groups, accessibility measures have also been used to examine
the equity of the transport and land use interaction (Bocarejo & Oviedo, 2012; Delmelle & Casas,
40
2012; Foth et al., 2013; Golub & Martens, 2014; Guzman et al., 2017). However, even though the
connection between transport and economic development has been extensively investigated,
insufficient research has coupled comprehensive measures of accessibility with urban
development.
Accessibility is increasingly being incorporated into metropolitan transport plans and national
planning guidelines, although mobility-planning remains the dominant paradigm (Boisjoly & El-
Geneidy, 2017c; D. Proffitt et al., 2017). In the United Kingdom, a national accessibility framework
exists, but analysis is still “generally too transport focused” and accessibility indicators are
“misused” and “abused” (COST, 2012; Halden, 2011). At the municipal or regional scale, cities
such as London, Paris, Sydney, and Atlanta are now employing the concept of accessibility, either
as an independent goal or objective, or as part of an environmental justice assessment (Boisjoly
& El-Geneidy, 2017a). In both Sydney and London, improving access to jobs or employment is
mentioned as a key method to support regional economic development, and the ‘30-minute’ city
is a key element to Sydney’s long-range plan (Greater Sydney Commission, 2018; NSW
Government, 2012; Transport for London, 2006). Canadian cities, however, have been slow to
adopt the concept; while their plans mention access to transit, only the discussion paper for the
updated “Big Move” for Toronto contains a metric for access to jobs by transit, with goals similar
to the London plan (Metrolinx, 2016). Similarly, in the United States, only a few cities have adopted
accessibility goals and performance metrics in their regional transport plans (D. Proffitt et al.,
2017). Accessibility planning practice thus remains limited across North America.
1.2 Transport, accessibility and urban development
A large body of literature has focused on establishing a theoretical framework between transport
and subsequent land use patterns and urban development. Kain (1962) and later Alonso (1964)
extended the model developed by von Thünen representing land value as a function of distance
to a central business district, and argued that land values in turn influence land use patterns. The
bid-rent theory developed by Alonso (1964), and later extended by many other scholars (see for
example Anas and Moses (1977); Mills (1967)), offers households a trade-off between transport
cost and rent, resulting in higher land values for more central locations. The area with the highest
accessibility attracts the most development and becomes the CBD. In a self-reinforcing process,
because it has greater access to both jobs and workers, competition will favor more intensive
development in this central location, and according to bid-rent theory, prices will rise. Changes in
the transport system are therefore said to result in changes in land use patterns through the
intermediating effect of commute duration and land values.
41
In a similar vein as the urban economics scholars before them, transport researchers focusing on
accessibility have linked transport changes to changing land use and activity patterns (El-Geneidy
& Levinson, 2006; Forkenbrock et al., 2001; Giuliano, 2004). Many governments and transit
agencies have also acknowledged the link between transport and economic development
(European Commision, 2010), and many cities and regions worldwide are looking to capitalize on
this link through land value capture (Medda, 2012; Salon & Shewmake, 2014; Smolka, 2013;
Transport for London, 2017; Zhao et al., 2012).
Public sector policy, and economic and population growth, play vital roles in determining the
viability of the links presented above (Giuliano, 2004; Warade, 2007). Supporting tax and land use
policies, for example, can expedite how changes in accessibility impact land use, while the general
economic climate is a vital aspect in determining whether or not development will occur on the
site. Banister and Berechman (2000) argue that coordination between regional and municipal
agencies, combined with favorable economic circumstances, are pre-conditions for the
association between transport and development to occur.
The links presented above have subsequently been investigated in a myriad of empirical studies.
Levinson (1998) examines the association between accessibility measures and commute
duration. In a cross-sectional study, he finds that, for origins, access to employment opportunities
is inversely related to average commute duration, while access to housing is positively correlated
to average commute time. The association between accessibility and land values is considered
by El-Geneidy, van Lierop, et al. (2016), Franklin and Waddell (2003) and Martínez and Viegas
(2009), among others, who find that higher accessibility levels are related to increased home
values. Iacono and Levinson (2015), on the other hand, conclude that, although homes in
neighbourhoods with higher accessibility levels command value premiums, the relationship no
longer holds for changes in accessibility. Maturity of the transport network is said to be causing
this effect. Similarly, Du and Mulley (2006) find that the effects of accessibility on home values
depend on location and the accessibility level of the neighbourhood.
The relationship between transport investments and economic benefits is assessed by Banister
and Berechman (2000), Mejia-Dorantes et al. (2012), and Padeiro (2013), among others. They
find that transport infrastructure changes are related to economic development, although the
relationship varies by location and occurs mostly in sectors showing large agglomeration
economies, such as finance and real estate. Mejia-Dorantes et al. (2012) show that distance to
subway stations is a key determinant of firm location, while Padeiro (2013), in a case study of
small municipalities in the Île-de-France region, concludes that the presence of train stations does
42
not significantly affect job growth, whereas the presence of a highway is only a significant predictor
of growth for the smallest municipalities.
Ozbay et al. (2003) investigate the relationship between accessibility and economic development
in the New York – New Jersey region and find that accessibility changes are related to changes
in employment growth (and therefore land use). Similarly, Alstadt et al. (2012) find that local
accessibility calculated at a 40-minute threshold is a strong factor impacting economic activity in
the service sector, while regional accessibility computed with a 3-hour threshold is more valued
by the manufacturing sector. In a case study of motorways in Portugal, Holl (2004) develops a
measure of market access similar to a gravity-based measure of access to the labour force, and
concludes that the odds of firm birth are higher for several manufacturing and construction sectors
when market access is larger. Applied to a case study in Chicago, Warade (2007) develops a
quasi-integrated land use and transport model and concludes that higher access to jobs is
associated with increased household density, whereas higher access to workers is related to
increased job density. Y. Shen et al. (2014) examine the effects of local and regional accessibility
on development near the Atocha station in Madrid, Spain. The authors find that accessibility, at
both the city and country level, is a significant predictor in determining land cover change.
Farber and Grandez Marino (2017) acknowledge the strong association between accessibility and
urban development, and generate a typology of planned stations in the Greater Toronto and
Hamilton Area based on development potential around the station and the projected change in
accessibility. The authors conclude that there exists considerable mismatch between development
potential and large predicted accessibility changes from planned stations. This conclusion
highlights the need for accessibility considerations when investing considerable amounts in new
transport infrastructure, in order to realize the full benefits of the planned investment. We contend
that the introduction of AOD can greatly benefit this process.
2. HYPOTHESES, DATA, AND METHODOLOGY
2.1 Hypotheses
Based on the theoretical accessibility and development framework and the empirical literature
presented above, we hypothesise that accessibility drives the following natural phenomena:
Hypothesis 1: Residents in neighbourhoods with high access to employment opportunities and
low access to workers experience the shortest average commute duration, and vice versa.
43
Hypothesis 2: Neighbourhoods with high accessibility levels to both employment opportunities and
workers attract more development.
• Hypothesis 2a: High access to workers draws in more businesses.
• Hypothesis 2b: High access to employment opportunities invites residential and
commercial development by influencing home location choice and leveraging
agglomeration economies.
TABLE 5 - Summary of AOD hypotheses
Areas with: High Access to Jobs Low Access to Jobs
High Access to Labour [Urban centre]
H2a: Attracts employment and residences (commercial and residential development)
H1a: Residents have long commutes
H2a: Attracts employment (commercial development)
Low Access to Labour H1b: Residents have short commutes
H2b: Attracts residences (residential development)
[Urban fringe]
Accessibility-oriented development invites planners to leverage access to steer, slow down, or
speed up the phenomena that naturally follow from accessibility changes as presented in the
hypotheses above. We therefore define accessibility-oriented development as a strategy that
balances accessibility between employment opportunities and workers in order to foster an
environment conducive to development. This differs from the traditional ‘jobs-housing balance’
literature by avoiding the use of arbitrary municipal boundaries and instead considers the
relationship between access to jobs and access to competing workers (Cervero, 1989, 1996;
Levinson, 1998; Levinson et al., 2017).
AOD will allow planners to leverage tools such as zoning measures to attract desired development
into TOD sites. For example, assume that an area has been designated as a TOD site in planning
documents. The zone can then become more attractive for commercial/industrial development –
if the hypotheses above hold – by increasing access to the labour force, which can be achieved
by (1) zoning a part of the proposed TOD as residential, to provide a direct customer/labour base,
and (2) offering new and improving current transport options to a variety of residential
neighbourhoods. AOD thus asks planners to leverage accessibility levels (in this case, increasing
44
access to workers) to attract desired development (commercial and industrial in the case of the
example).
In short, employing AOD in focus areas by increasing job accessibility should help shorten
commute times and attract residents to these neighbourhoods, helping these regions to
rejuvenate. Other areas could be designed to allow for maximum access to the labour force, which
would provide incentives for firms in the service and retail sectors to locate themselves in these
neighbourhoods, in order to minimize their employees’ or customers’ travel times and benefit from
agglomeration economies. This in turn would lower commute times. Development would therefore
occur naturally in sectors planned with AOD principles, once the starting conditions are set by
adequate policy.
Ideally, TOD can be understood as a component of AOD. Whereas TOD solely focuses on local
access/egress conditions to public transport and the local distribution of land use, AOD also
considers access by public transport and the regional distribution of land use, as well as access
to destinations by other modes.
2.2 Case study
To test the hypotheses about the phenomena that naturally follow from accessibility levels and
thereby the validity of using an AOD approach, a case study is performed in the Greater Toronto
and Hamilton Area, Canada (GTHA) between 2001 and 2011. The GTHA is the largest
metropolitan agglomeration in Canada, housing 6.6 million residents in 2011 and comprises the
Hamilton, Toronto and Oshawa census metropolitan areas (CMAs). Population in the region
increased by over 1 million inhabitants during the study period, while the total number of jobs grew
from 2.9 to 3.5 million (Statistics Canada, 2015). Between 2001 and 2011, the transport network
in the region underwent substantial changes: a new subway line was opened in 2002, and several
new train stations were constructed. A context map of the GTHA can be seen in figure 10.
Note that the choice of this particular case study is largely irrelevant in providing the foundations
for an AOD approach. The case study only serves to corroborate that the two hypotheses about
accessibility hold, even in a region where accessibility-oriented development tactics have not been
employed. If, and only if, the hypotheses hold, AOD will be a valid strategy to develop sites or
neighbourhoods.
45
FIGURE 10 - Context map of the Greater Toronto and Hamilton Area
2.3 Data and Methods
To generate cumulative accessibility measures by both car and public transport (PT), data from
Metrolinx and Statistics Canada were used. The cumulative opportunities metric was chosen
because it is easier to communicate to planners and urban decision-makers, increasing its
chances to impact policy (Geurs & van Wee, 2004; Handy & Niemeier, 1997). First, travel times
between census tract centroids, modeled through the EMME software, by both car and public
transport in 2001 were acquired from Metrolinx, in addition to car travel times in 2011. Public
transport travel times in 2011 were subsequently calculated by making use of data from the
General Transit Feed Specification (GTFS). A network was built based on this data in ArcGIS
through the tool ‘Add GTFS to a network dataset’, and fastest routes were calculated between
each census tract centroid on a regular Tuesday at 8 AM in 2011. The fastest path algorithm took
into account walking time, waiting time, and in-vehicle time (as determined by the transit
46
schedule), but did not assign a penalty for transfers. Census data provides the number of jobs and
the population in the labour force in each census tract.
To reflect the commuting behavior of an average individual, car accessibility was calculated for a
time threshold of 30 minutes (equivalent to the average commute duration in the region by car),
while access by transit was computed for a 45 minute time threshold (equivalent to the average
commute duration in the region by public transport (Statistics Canada, 2010)). A cumulative
measure of accessibility then counts the number of opportunities that can be reached within that
threshold. As the data sources for the number of jobs differed between 2001 and 2011, a relative
measure of accessibility was calculated by dividing the total number of jobs (workers) reachable
within the threshold by the total number of jobs (workers) in the region. Accessibility can then be
interpreted as the percentage of all jobs (workers) in the region an individual can access: a value
of 1 signifies that all jobs (workers) can be reached within the threshold (30 or 45 minutes
depending on the mode), while a value of 0.25 indicates that 25% of all jobs (workers) can be
reached within the set time frame. The accessibility calculations were performed for each census
tract in the GTHA.
To test the two AOD hypotheses, commute duration for 2011 was gathered from Statistics
Canada, while commute duration in 2001 was calculated based on travel times and origin-
destination flows provided by Statistics Canada. Development was subsequently measured by the
change in the percentage of open area in the census tract (measured as the area not used for
residential, commercial, industrial, governmental, or park purposes). The AOD hypotheses were
then examined through five regression models, relating commute duration, open area, and job
and population density with accessibility and accessibility changes.
A first, cross-sectional, model predicts average commute duration in 2001 based on accessibility
in 2001 and a dummy variable for the Hamilton CMA. A dummy variable for Hamilton was
introduced to reflect that residents of census tracts in the Hamilton CMA are more likely to
commute to Hamilton than Toronto, thus their commute time is, on average, lower than in the
Toronto or Oshawa census metropolitan areas.
A second model, to test if the relationship between commute duration and accessibility also holds
over time, predicts commute time in 2011 based on observed commute times in 2001 (by car and
public transport combined) and accessibility in 2001, as well as changes in accessibility levels
between 2001 and 2011. Levels of accessibility in 2001 were included as it is assumed that the
initial situation will influence how changes occur (Putnam, 1983). Model 1 and 2 together thus
47
examine the link between accessibility and commuting behavior, in order to validate our first AOD
hypothesis, namely that inhabitants of AOD areas with high access to jobs and low access to the
labour force experience the lowest commute times.
A third model was developed to assess the second AOD hypothesis, with open area
acting as a proxy for development. Open area was measured as the area not used for residential,
commercial, industrial, governmental, or park purposes. The same model specification as the
second model is used: open area in 2011 is predicted based on open area and accessibility in
2001, and changes in accessibility between the two years. In order to disentangle the separate
effects of labour and employment accessibility on attracting residential, commercial, and industrial
development, two extra regressions were performed: one predicting job density and the other
predicting population density in 2011. Descriptive statistics of the variables used in the different
models are shown in Table 6.
TABLE 6 - Descriptive statistics for commute duration, accessibility and development data
Variable Description Mean Standard dev.
Commute01 Average commute time in 2001 (min) (all modes) 28.58 6.55 Commute11 Average commute time in 2011 (min) (all modes) 31.29 4.25 Access01 to Jobs by Car Access to jobs by car in 30 minutes in 2001 (%) 20.12 12.25 Access01 to Workers by Car Access to workers by car in 30 minutes in 2001 (%) 20.75 6.86 Access01 to Jobs by PT Access to jobs by PT in 45 minutes in 2001 (%) 8.90 9.97 Access01 to Workers by PT Access to workers by PT in 45 minutes in 2001 (%) 7.53 6.79 Access11 to Jobs by Car Access to jobs by car in 30 minutes in 2011 (%) 30.97 20.24 Access11 to Workers by Car Access to workers by car in 30 minutes in 2011 (%) 29.26 10.36 Access11 to Jobs by PT Access to jobs by PT in 45 minutes in 2011 (%) 8.17 8.62 Access11 to Workers by PT Access to workers by PT in 45 minutes in 2011 (%) 6.32 5.20 ∆ Commute Change in commute time (min) (all modes) 2.71 4.70
∆ Access to Jobs Car Change in access to jobs by car (%) 10.84 10.17
∆ Access to Workers by Car Change in access to workers by car (%) 16.74 7.59 ∆ Access to Jobs by PT Change in access to jobs by PT (%) -0.73 3.34 ∆ Access to Workers by PT Change in access to workers by PT (%) -1.21 2.67 OpenArea01 Percentage of open area in 2001 (%) 14.61 15.92 OpenArea11 Percentage of open area in 2011 (%) 14.57 24.13 JobDens01 Job density in 2001 (jobs/km2) 181.45 526.07 JobDens11 Job density in 2011 (jobs/km2) 164.64 544.94 PopDens01 Population density in 2001 (population/km2) 4337.34 4781.05 PopDens11 Population density in 2011 (population/km2) 4903.45 5285.02
48
3. RESULTS AND DISCUSSION
Figure 11 shows normalized accessibility levels by car to employment opportunities and workers.
Access to jobs by car is highest in downtown Toronto, while the highest accessibility levels to
workers are present in neighbourhoods that form a ring around the Toronto CBD. This reflects that
the central business district houses fewer people than the area immediately surrounding it, and
that it is easier for residents of the outskirts of the region to travel to these suburban locations than
to the city centre. Between 2001 and 2011, access to workers increased substantially more across
the study area than access to jobs. According to the second AOD hypothesis, the suburban
locations with high access to the labour force should experience more job creation during the
study period, providing that the benefits of access to labour outweigh those of existing
agglomeration economies of access to existing businesses (operationalized as access to jobs).
Accessibility levels by public transport are shown in figure 12. Access by transit is considerably
lower than access by car, even with an extra 15 minutes of travel time, in both years, for access
to jobs and workers. High access by transit is mainly present in downtown Toronto and in areas
located in close proximity to the commuter rail lines. Unlike the spatial patterns present in access
by car, the two accessibility measures for public transport, to jobs and workers, are highly
correlated (a correlation of 0.95). In 2011, suburban areas located next to the public transport
network have seen increases in accessibility, while areas with traditionally high access (such as
downtown Toronto) have seen a small decrease in access, which might be related to an ongoing
suburbanization of jobs, combined with investments made in the commuter train network during
the study period.
49
FIGURE 11 - Access to jobs and the labour force by car within 30 minutes
50
FIGURE 12 - Access to jobs and the labour force by public transport within 45 minutes
3.1 Accessibility, commute duration and urban development
The results of the model associating average commute duration in 2001 and accessibility (Model
1) are shown in Table 7. Note that access by public transport was not included in this model due
to collinearity with access by car. A separate model was tested for public transport accessibility
and resulted in similar conclusions, but was excluded from the analysis due to its similarity with
the reported model.
Higher access to jobs is related to shorter commute times, ceteris paribus, while a higher access
to the labour force is related to longer commute times, all else equal, which is consistent with the
51
findings from Levinson (1998). In absolute terms, an extra 100,000 accessible jobs decreases
commute time by 0.48 minutes (-0.14 minutes per percent), while an extra 100,000 workers
increases average commute duration by 0.87 minutes (0.27 minutes per percent). These results
corroborate the first AOD hypothesis: AOD areas with high access to jobs and low access to the
labour force have shorter average commute times than the rest of the region (and vice versa).
The dummy variable for Hamilton indicates that, all else equal, commute time in the Hamilton
census metropolitan area is 6.4 minutes shorter. Note that accessibility levels also influence the
predicted commute duration in Hamilton. Evaluated at the average accessibility levels for Hamilton
(9% of all jobs accessible by car and 12% of all workers accessible by car), census tracts in
Hamilton have an average predicted commute duration of 22.2 minutes, 7.2 minutes less than the
predicted average for the entire study area.
TABLE 7 - Regression model predicting average commute duration in 2001
Variable Coefficient Sig. Confidence int.† Hypothesis
Intercept 26.5799 *** [25.3558, 27.8040]
Access01 to Jobs by Car -0.1411 *** [-0.1699, -0.1124] -
Access01 to Workers by Car 0.2722 *** [0.2178, 0.3265] +
Hamilton dummy -6.3950 *** [-7.4930, -5.2971] -
Adjusted R2 0.2212
Dependent Variable: Average commute duration in 2001 * 95% significance level | ** 99% significance level | *** 99.9% significance level † 95% confidence interval
The results of the temporal model relating commute time and accessibility (Model 2) are shown in
Table 8. Almost 60% of the total variation in commute times in 2011 is explained by this model.
The coefficients for accessibility in 2001 have the expected signs and statistical significance:
access to jobs in 2001 is associated with a shorter commute time, while access to the labour force
is related to a longer commute duration. The statistical significance of both coefficients could be
related to a time lag between accessibility levels and commute patterns adjusting themselves to
the new situation, i.e., commute patterns in 2001 were not yet in equilibrium with respect to 2001
accessibility.
Unlike in the cross-sectional model, the effects of both changes in access by public transport and
car can be investigated separately, as their changes are no longer correlated. Notably, only
changes in access to workers by car and access to jobs by public transport are statistically
significant predictors of average commute duration in 2011. These results confirm that the first
AOD hypothesis also holds over time. An increase in the change in accessibility of 1% of all
52
workers by car lengthens commutes by 0.2 minutes, while a 1% increase in access by public
transport to all jobs reduces commutes by 0.1 minutes. Interestingly, the relative magnitudes of
both coefficients are reversed compared to the cross-sectional model.
The two coefficients for change in access to jobs by car and to workers by public transport were
found to be not statistically significant. We hypothesize that this is related to the maturity of the
transport network in the region. Small changes to the network can no longer induce large impacts
on accessibility levels (Gómez-Ibáñez, 1985), resulting in diminishing returns to the outcomes of
transport investments (Iacono & Levinson, 2015).
TABLE 8 - Commute time and open area in 2011 fitted to accessibility in 2001 and changes in accessibility between 2001 and 2011
Commute duration in 2011 Open area in 2011
Variable Coeff. Sig Confidence int. † Hyp. Coeff. Sig Confidence int. † Hyp.
Intercept 17.8976 *** [17.0579, 18.7372] 21.3201 *** [17.9573, 24.6829] Access01 to Jobs by Car -0.1027 *** [-0.1250, -0.0804] - -0.1948 ** [-0.3185, -0.0710] - Access01 to Workers by Car 0.0793 *** [0.04652, 0.1122] + -0.6212 *** [-0.8162, -0.4262] - ∆ Access to Jobs Car -0.0093 [-0.0306, 0.0121] - 0.0251 [-0.0989, 0.1490] - ∆ Access to Workers by Car 0.2139 *** [0.1745, 0.2532] + -0.0595 [-0.2875, 0.1685] - ∆ Access to Jobs by PT -0.1083 *** [-0.1713, -0.0453] - 0.2876 [-0.0810, 0.6561] - ∆ Access to Workers by PT -0.0652 [-0.1558, 0.0254] + -0.6645 * [-1.1883, -0.14075] - Commute01 (Observed) 0.3561 *** [0.3295, 0.3826] + OpenArea01 0.4958 *** [0.4633, 0.5282] +
Adjusted R2 0.5922 0.5459 Dependent Variables: Average commute duration and open area in 2011 * 95% significance level | ** 99% significance level | *** 99.9% significance level † 95% confidence interval The results for the model predicting the percentage of open area in each census tract in 2011
(Model 3) can be seen in Table 8, explaining 55% of all variation in open space. The statistically
significant coefficients for accessibility in 2001 corroborate the second AOD hypothesis: access
to jobs and workers in 2001 are associated with decreases in open area. One extra percent of
access to jobs by car in 2001 reduces open space by 0.19%, and an extra percent of access to
workers decreases open space 0.62%. Residential, commercial, and industrial development thus
seems to be associated with AOD areas.
Changes in accessibility levels, except for the change in worker access by public transport, are
not statistically significant predictors of open space in 2011. Two explanations are possible. First,
location choices do not occur often due to the associated capital costs, thus there exists a
substantial time lag between accessibility levels changing and location choice. A study period
encompassing only 10 years will therefore not be able to fully capture these long-term decisions,
especially as it is unknown when each accessibility change occurred. It is also expected that firms,
53
rather than individuals, are more sensitive to changes in access to jobs and workers, and are more
prone to change their locations (as residents also place high value on access to other
opportunities, such as schools, shops, and social networks). The statistically significant coefficient
for the change in access to workers by transit corroborates this, as it is expected that access to
workers is an attractor in firm location behavior. As only the change in access to workers by public
transport is statistically significant, we can conclude that firms in the GTHA are more likely to
locate near areas where transit service, instead of car accessibility, increases. Although this
relationship might depend on the business sector and their associated transport costs for their
products and employees, it could be indicative of a paradigm shift in the way (some) enterprises
expect their employees or customers to travel. Second, as some areas are almost fully built,
changes in accessibility in these neighbourhoods can no longer reduce open space and can
therefore not be captured by the model.
To resolve this second possibility, and to confirm the hypotheses about firm and individual
behavior mentioned above (hypotheses 2a and 2b), two extra models were computed, predicting
job and population density in 2011, see Table 9. These models again confirm the second AOD
hypothesis: job density increased more in areas where baseline access to workers was highest,
whereas population density grew considerably more in areas where 2001 access to jobs was
highest. This corroborates the hypothesis that firms are attracted to where workers and customers
are located, whereas individuals are more likely to choose a home with high access to job
opportunities.
Surprisingly, access to jobs in 2001 is not significant in the model predicting job density and has
a negative coefficient, indicating that businesses were more likely to locate away from existing
jobs. When a squared term of this variable is added to the model, the relationship follows a more
intuitive pattern, although it is still insignificant: once there is critical mass of job accessibility,
businesses are attracted to job-rich areas, corroborating the importance of agglomeration
economies. Furthermore, this research uses a crude measure of ‘jobs’, implicitly assuming all jobs
are equivalent. While some jobs are in agglomeration-favoring industries (like finance and
government), others need to be near, but not central to, agglomerations (like industry), or near
customers (retail, services).
Among the change variables, only the change in access to the labour force by transit is statistically
significant: a 1 percent increase in access to workers by public transport between 2001 and 2011
is associated with an extra 8 jobs per square kilometer. As with the model predicting open area, it
54
appears that firms in the region were more likely to be attracted to areas where the public transport
system, instead of the highway and street network, improved.
The model predicting population density in 2011 shows that all changes in accessibility, except for
the change in access to jobs by transit, are statistically significant. The two coefficients for the
change in access to the labour force by car and transit are positive, suggesting that individuals
are attracted to locations where worker accessibility increased during the study period. Note that
the significance of these variables could be related to reverse causality: as job-rich areas attract
more residents, worker accessibility will necessarily increase in and surrounding these areas. The
change in population density in the neighbourhood and in surrounding census tracts might
therefore cause the change in worker access. The coefficient for the change in access to jobs by
car is negative and thus contradicts our hypothesis: an increase in job accessibility of 1% between
2001 and 2011 is related to a decrease in population density of 23 inhabitants per km2. This might,
however, indicate a trade-off between residential and commercial development in a census tract,
or might be related to larger scale zoning patterns that do not allow concurrent residential and
commercial development in a single zone. Nevertheless, the significance and signs of the majority
of the variables in the model corroborate the second hypothesis.
TABLE 9 - Job and population density in 2011 fitted to accessibility in 2001 and changes in
accessibility between 2001 and 2011
Job density in 2011 Population density in 2011
Variable Coeff. Sig Confidence int. † Hyp. Coeff. Sig Confidence int. † Hyp.
Intercept -31.9827 * [-62.2608, -1.7045] 81.3165 [-379.6737, 542.3068] Access01 to Jobs by Car -1.2361 [-2.5445, 0.0724] 21.3944 * [ 1 .9579 , 40 .8308 ] + Access01 to Workers by Car 2.1506 * [0.1992, 4.1020] + -21.9910 [-51.3465, 7.3645] ∆ Access to Jobs Car 0.4887 [-0.7560, 1.7333] -23.0279 * [-41.8525, -4.2032] + ∆ Access to Workers by Car -0.0960 [-2.3794, 2.1874] + 67.2478 *** [32.6678, 101.8279] ∆ Access to Jobs by PT -1.6380 [-5.3603, 2.0844] -37.1115 [-93.4974, 19.2743] + ∆ Access to Workers by PT 8.4400 ** [3.2029, 13.6772] + 174.8577 *** [95.6686, 254.0467] JobDensity01 1.0042 *** [0.9853, 1.0231] + PopDensity01 0.9584 *** [ 0 . 9 2 5 9 , 0 . 9 9 0 9 ] +
Adjusted R2 0.931 0.787
Dependent Variables: Average commute duration and open area in 2011 * 95% significance level | ** 99% significance level | *** 99.9% significance level † 95% confidence interval
4. CONCLUSION
Accessibility-oriented development, both a market process and a strategy that aims to balance
accessibility between employment opportunities and workers in order to foster an environment
conducive to development, has been shown to be associated with changing commute times and
55
economic development. Through AOD planners can leverage the relationship between transport
and land use patterns by explicitly considering the functional connections between local and
regional transport investments and local and regional land use. In contrast with TOD, AOD also
considers the access that is provided by transit, instead of only examining access to transit.
The regression models in our study indicate that, through AOD, planners can capitalize on two
tangible benefits that accrue to neighbourhoods when accessibility levels change. First, by
influencing access to jobs and/or workers through land use (e.g. zoning) or transport changes,
average commute times can be adjusted across neighbourhoods: increases in access to jobs are
related to decreases in commute duration, while increases in access to workers are associated
with longer average commute times. Second, higher access to jobs and/or workers is associated
with residential, commercial, and industrial development.
It is important to note that the relationships uncovered in this study are not conclusive, nor can
they determine a causal relationship; more studies would need to be developed in multiple cities
to further corroborate these findings. Furthermore, the analyses conducted in this study were
performed under the assumption that the land market in the GTHA operates in perfect market
conditions. Toronto, however, as with most cities in the world, regulates and prioritizes certain
land uses in their many plans and programs, potentially altering the effects of accessibility on
location choices in favor of the city’s development guidelines.
Nevertheless, this study provides strong evidence of the relationship between accessibility,
commute duration, and residential and firm location. Planners and urban decision-makers aiming
to create successful developments should therefore not limit themselves to using site-specific
planning tools such as TOD, but should also take into account AOD principles, by measuring the
impacts of new transport or land use plans in terms of access to both employment opportunities
and workers, and then leveraging these accessibility levels to generate desired development.
ACKNOWLEDGEMENTS
This research was funded by the Social Sciences and Humanities Research Council of Canada
and the Natural Sciences and Engineering Research Council of Canada. We would like to thank
David King, Nicholas Day and Joshua Engel-Yan from Metrolinx for providing travel time data.
56
CONCLUSION
Accessibility is increasingly being used in both research and practice as a key performance
indicator for integrated land use and transport planning. Many cities around the world, in their
respective transport plans, have now adopted the concept and intend to increase access to
destinations to reduce social exclusion and provide a catalyst for economic growth.
Among Canada’s eleven largest metropolitan areas, Toronto residents experience the highest
access to jobs, followed by those in Montréal, Vancouver, Calgary and Ottawa. When examining
access to low-income jobs, however, vulnerable inhabitants of Montréal are much better off than
in any other Canadian city. Overall, vulnerable individuals experience higher accessibility levels
than their fellow residents in their respective cities, indicating that accessibility is equitably
distributed in Canada’s largest metropolitan areas. However, poverty is slowly being suburbanized
as city centres and neighbourhoods next to transit hubs are increasingly being re-developed and
re-vitalized. The most vulnerable households might therefore be pressured to move to areas
where accessibility is (much) lower than in their previous locations, which could negatively affect
vertical equity in many of Canada’s large cities.
Government policies intending to increase accessibility in socially-vulnerable neighbourhoods can
bring considerable benefits to these areas. The results from chapter 2 suggest that, for low and
medium income census tracts, increases in transit accessibility are related to higher increases in
income. An improvement in accessibility to jobs by car is similarly associated with larger income
increases. Furthermore, unemployment rates rose less in areas where accessibility levels
improved. The equity of accessibility to employment opportunities thus plays a key role in
determining resulting equity of outcome, stressing the need for methods that can incorporate
equity considerations into the evaluation of new transport projects.
The results from chapter 3 suggest that accessibility improvements can also create considerable
economic benefits: increases in accessibility to jobs and/or workers in the Greater Toronto and
57
Hamilton Area were associated with more residential, commercial, and industrial development.
Moreover, accessibility levels influenced commute durations: high accessibility to jobs was related
to shorter commutes and higher accessibility to people was associated with longer commutes.
By embracing policies to increase accessibility levels, cities can thus improve quality of life for
their most vulnerable inhabitants while at the same time attracting economic development. Cities
and their transport planning agencies can therefore greatly benefit from adopting accessibility
indicators and incorporating these into their plans and policies.
58
REFERENCES
Ahlfeldt, G. M., & Wendland, N. (2011). Fifty years of urban accessibility: The impact of the urban
railway network on the land gradient in Berlin 1890–1936. Regional Science and Urban
Economics, 41(2), 77 - 88.
Alonso, W. (1964). Location and Land Use. Cambridge, MA: Harvard University Press.
Alstadt, B., Weisbrod, G., & Cutler, D. (2012). The relationship of transportation access and
connectivity to local economic outcomes: A statistical analysis. Transportation Research
Record: Journal of the Transportation Research Board(2297), 154 - 162.
Anas, A., & Moses, L. (1977). Mode choice, transport structure and urban land use. Journal of
Urban Economics, 6.
Andersson, F., Haltiwanger, J., Kutzbach, M., Pollakowski, H., & Weinberg, D. (2014). Job
displacement and the duration of joblessness: The role of spatial mismatch. National
Bureau of Economic Research. Cambridge, MA.
Atlanta Regional Commission. (2016). The Atlanta region's plan: Transportation.
Banister, D., & Berechman, J. (2000). Transport investment and economic development. London,
UK: UCL Press.
Barker, T. C., & Robbins, M. (1963). A history of London transport: Volume I - The nineteenth
century. London: Allen & Unwin.
Belzer, D., & Autler, G. (2002). Transit oriented development: Moving from rhetoric to reality.
Retrieved from Washington, DC:
Bertolini, L., Curtis, C., & Renne, J. (2012). Station area projects in Europe and beyond: Towards
transit oriented development? Built Environment, 38(1), 31 - 50.
Bissonnette, L., Wilson, K., Bell, S., & Shah, T. I. (2012). Neighbourhoods and potential access to
health care: The role of spatial and aspatial factors. Health & Place, 18, 841 - 853.
59
Blumenberg, E., & Pierce, G. (2014). A Driving Factor in Mobility? Transportation's Role in
Connecting Subsidized Housing and Employment Outcomes in the Moving to Opportunity
(MTO) Program. Journal of the American Planning Association, 80(1), 52 - 66.
Blumenberg, E., & Pierce, G. (2017). The drive to work: The relationship between transportation
access, housing assistance, and employment among participants in the Welfare to Work
Voucher Program. Journal of Planning Education and Research, 37(1), 66 - 82.
Boarnet, M. (2011). A broader context for land use and travel behavior, and a research agenda.
Journal of the American Planning Association, 77(3), 197 - 213.
Bocarejo, J., & Oviedo, D. (2012). Transport accessibility and social inequities: A tool for
identification of mobility needs and evaluation of transport investments. Journal of
Transport Geography, 24, 142-154.
Boisjoly, G., & El-Geneidy, A. (2017a). How to get there? A critical assessment of accessibility
objectives and indicators in metropolitan transportation plans. Transport Policy, 55, 38-50.
Boisjoly, G., & El-Geneidy, A. (2017b). How to get there? A critical assessment of accessibility
objectives and indicators in metropolitan transportation plans. Transport Policy.
Boisjoly, G., & El-Geneidy, A. (2017c). The insider: A planners’ perspective on accessibility.
Transport Geography, 64, 33 - 43.
Calthorpe, P. (1993). The next American metropolis: Ecology, community, and the American
Dream. New York: Princeton Architectural Press.
Cervero, R. (1989). Jobs-housing balancing and regional mobility. Journal of the American
Planning Association, 55(2), 136 - 150.
Cervero, R. (1996). Jobs-housing balance revisited: trends and impacts in the San Francisco Bay
Area. Journal of the American Planning Association, 62(4), 492 - 511.
Cervero, R. (1997). Paradigm shift: From automobility to accessibility planning. Urban Futures(22),
9 - 20.
Cervero, R., Christopher, F., & Murphy, S. (2002). Transit-oriented development and joint
development in the United States: A literature review. Transportation Research Board,
National Research Council. Washington, DC.
Cervero, R., Murphy, S., Ferrell, C., Goguts, N., Arrington, G. B., Boroski, J., . . . Witenstein, N.
(2004). Transit oriented development in America: experiences, challenges, and prospects
(Vol. TCRP Report 102). Washington, DC: National Academy Press.
Chatman, D. (2006). Transit-oriented development and household travel: A study of California
cities. Retrieved from Los Angeles, CA: https://trid.trb.org/view.aspx?id=864200
60
Chatman, D. (2013). Does TOD Need the T? Journal of the American Planning Association, 79(1),
17 - 31.
City of Denver. (2014). Transit oriented development strategic plan. Retrieved from
https://www.denvergov.org/content/dam/denvergov/Portals/193/documents/TOD_Plan/T
OD_Strategic_Plan_FINAL.pdf
City of Edmonton. (2009). The way we move: Transportation master plan.
City of Ottawa. (2013). Transportation master plan.
Congress for the New Urbanism. (1999). Charter of The New Urbanism: McGraw-Hill.
Conseil régional d'Île‐de‐France. (2014). Plan de déplacements urbains Île‐de‐France.
COST. (2012). Accessibility instruments for planning practice. Retrieved from
http://www.accessibilityplanning.eu/wp-content/uploads/2012/10/COST-Report-1-
FINAL.pdf
Curtis, C., Renne, J., & Bertolini, L. (2009). Transit-oriented development: Making it happen.
Burlington, VT: Ashgate.
Dalvi, M. Q., & Martin, K. M. (1976). The measurement of accessibility: Some preliminary results.
Transportation, 5, 17 - 42.
Damant-Sirois, G., Grimsrud, M., & El-Geneidy, A. (2014). What’s your type: A multidimensional
cyclist typology. Transportation, 41(6), 1153 - 1169.
Delbosc, A., & Currie, G. (2011). Using Lorenz curves to assess public transport equity. Journal
of Transport Geography, 19(6), 1252-1259. doi:10.1016/j.jtrangeo.2011.02.008
Delmelle, E., & Casas, I. (2012). Evaluating the spatial equity of bus rapid transit-based
accessibility patterns in a developing country: The case of Cali, Colombia. Transport
Policy, 20, 36-46. doi:10.1016/j.tranpol.2011.12.001
Du, H., & Mulley, C. (2006). Relationship between transport accessibility and land value.
Transportation Research Record(1977), 197 - 205.
Duncan, M. (2011). The impact of transit-oriented development on housing prices in San Diego,
CA. Urban Studies, 48(1), 101 -127.
Dyos, H. J. (1955). Railways and housing in Victorian London. The Journal of Transport History,
2, 90 - 100.
El-Geneidy, A., & Levinson, D. (2006). Access to destinations: Development of accessibility
measures. Retrieved from St-Paul, Minnesota, U.S.:
http://nexus.umn.edu/projects/Access/Access-FinalReport.pdf
61
El-Geneidy, A., Levinson, D., Diab, E., Boisjoly, G., Verbich, D., & Loong, C. (2016). The cost of
equity: Assessing transit accessibility and social disparity using total travel cost.
Transportation Research Part A: Policy and Practice, 91, 302 - 316.
El-Geneidy, A., van Lierop, D., & Wasfi, R. (2016). Do people value bicycle sharing? A multilevel
longitudinal analysis capturing the impact of bicycle sharing on residential sales in
Montreal, Canada. Transport Policy, 51, 174-181.
European Commision. (2010). High-speed Europe. Retrieved from
https://ec.europa.eu/transport/sites/transport/files/themes/infrastructure/studies/doc/2010
_high_speed_rail_en.pdf
Fan, Y., Guthrie, A., & Levinson, D. (2012). Impact of light rail implementation on labor market
accessibility: A transportation equity perspective. The Journal of Transport and Land Use,
5(3), 28 - 39.
Farber, S., & Grandez Marino, M. (2017). Transit accessibility, land development and
socioeconomic priority: A typology of planned station catchment areas in the Greater
Toronto and Hamilton Area. Journal of Transport and Land Use, 10(1), 33 - 56.
Forkenbrock, D., Mathur, S., & Schweitzer, L. (2001). Transportation investment and urban land
use patterns. Retrieved from
http://ir.uiowa.edu/cgi/viewcontent.cgi?article=1013&context=ppc_transportation
Foth, N., Manaugh, K., & El-Geneidy, A. (2013). Towards equitable transit: Examining transit
accessibility and social need in Toronto, Canada, 1996 - 2006. Journal of Transport
Geography, 29, 1 - 10.
Franklin, J., & Waddell, P. (2003). A hedonic regression of home prices in King County,
Washington, using activity-specific accessibility measures. Paper presented at the
Transportation Research Board 82nd Annual Meeting, Washington, DC.
Freeman, L. (2005). Displacement or succession? Residential mobility in gentrifying
neighborhoods. Urban Affairs Review, 40(4), 463 - 491.
Geurs, K., & van Wee, B. (2004). Accessibility evaluation of land-use and transport strategies:
Review and research directions. Journal of Transport Geography, 12(2), 127 - 140.
Gilat, M., & Sussman, J. (2003). Coordinated transportation and land use planning in the
developing world: Case of Mexico City. Transportation Research Record(1859), 102 - 109.
Giuliano, G. (2004). Land use impacts of transportation investments: Highway and Transit. In S.
Hanson & G. Giuliano (Eds.), The Geography of Urban Transportation. New York, NY: The
Guilford Press.
62
Golub, A., & Martens, K. (2014). Using principles of justice to assess the modal equity of regional
transportation plans. Journal of Transport Geography, 41, 10-20.
Gómez-Ibáñez, J. (1985). Transportation policy as a tool for shaping metropolitan development
Research in Transportation Economics, Volume 2 (pp. 55 - 81). Greenwich, CT: JAI Press.
Greater Sydney Commission. (2018). A metropolis of three cities: The Greater Sydney region plan
Retrieved from Sydney: https://gsc-public-1.s3.amazonaws.com/s3fs-public/greater-
sydney-region-plan-0318.pdf
Grengs, J. (2015). Nonwork accessibility as a social equity indicator. International Journal of
Sustainable Transportation, 9(1), 1 - 14.
Grube-Cavers, A., & Patterson, Z. (2015). Urban rapid rail transit and gentrification in Canadian
urban centres: A survival analysis approach. Urban Studies, 52(1), 178 - 194.
Guagliardo, M. F. (2004). Spatial accessibility of primary care: concepts, methods and challenges.
International Journal of Health Geographics, 3(3).
Guthrie, A., & Fan, Y. (2016). Developers’ perspectives on transit-oriented development.
Transport Policy, 51, 103 - 114.
Guzman, L., Oviedo, D., & Rivera, C. (2017). Assessing equity in transport accessibility to work
and study: The Bogotá region. Journal of Transport Geography, 58, 236 - 246.
Halden, D. (2011). The use and abuse of accessibility measures in UK passenger transport
planning. Research in Transportation Business & Management, 2, 12 - 19.
Handy, S. (2008). Regional transportation planning in the US: An examination of changes in
technical aspects of the planning process in response to changing goals. Transport Policy,
15(2), 113-126.
Handy, S., & Niemeier, D. (1997). Measuring accessibility: an exploration of issues and
alternatives. Environment and planning A, 29(7), 1175 - 1194.
Hansen, W. (1959). How accessibility shapes land use. Journal of the American Institute of
Planners, 25(2), 73 - 76.
Haywood, R. (1997). Railways, urban form and town planning in London: 1900-1947. Planning
Perspectives, 12(1), 37 - 69
Holl, A. (2004). Transport infrastructure, agglomeration economies, and firm birth: Empirical
evidence from Portugal. Journal of Regional Science, 44(4), 693 - 712.
Hu, L. (2017). Job accessibility and employment outcomes: Which income groups benefit the
most? Transportation, 44(6), 1421 - 1443.
Iacono, M., & Levinson, D. (2015). Accessibility dynamics and location premia: Do land values
follow accessibility changes? Urban Studies, 54(2), 364 - 381.
63
Ingram, D. R. (1971). The concept of accessibility: A search for an operational form. Regional
Studies, 5(2), 101 - 107.
Jacobson, J., & Forsyth, A. (2008). Seven American TODs: Good practices for urban design in
Transit-Oriented Development projects. Journal of Transport and Land Use, 1(2), 51-88.
Jensen, M. (1999). Passion and heart in transport — a sociological analysis on transport
behaviour. Transport Policy, 6(1), 19 - 33.
Kahn, M. (2007). Gentrification Trends in New Transit-Oriented Communities: Evidence from 14
Cities That Expanded and Built Rail Transit Systems. Real Estate Economics, 35(2), 155
- 182.
Kain, J. (1962). The journey-to-work as a determinant of residential location. Papers in Regional
Science, 9(1), 137 - 160.
Kamruzzaman, M., Shatu, F., Hine, J., & Turrell, G. (2015). Commuting mode choice in transit
oriented development: Disentangling the effects of competitive neighbourhoods, travel
attitudes, and self-selection. Transport Policy, 42, 187 - 196.
Karner, A., & Niemeier, D. (2013). Civil rights guidance and equity analysis methods for regional
transportation plans: A critical review of literature and practice. Journal of Transport
Geography, 33, 126 - 134.
Kawabata, M., & Shen, Q. (2007). Commuting inequality between cars and public transit: The
case of the San Francisco Bay Area, 1990-2000. Urban Studies, 44(9), 1759-1780.
Koenig, J. G. (1980). Indicators of urban accessibility: Theory and application. Transportation, 9,
145 - 172.
Korsu, E., & Wenglenski, S. (2010). Job accessibility, residential segregation, and risk of long-
term unemployment in the Paris region. Urban Studies, 47(11), 2279 - 2324.
Krizek, K., & El-Geneidy, A. (2007). Segmenting preferences and habits of transit users and non-
users. Journal of Public Transportation, 10(3), 71 - 94.
Langlois, M., van Lierop, D., Wasfi, R., & El-Geneidy, A. (2015). Chasing sustainability: Do new
TOD residents adopt more sustainable modes of transportation? Transportation Research
Record(2531), 83 - 92.
Larson, S. (2017). Social Equity and Public Transit: A Comparative Analysis of Persisting
Economic Outcomes and Accountability Indicators. University of Colorado at Denver.
Legrain, A., Buliung, R., & El-Geneidy, A. (2016). Travelling fair: Targeting equitable transit by
understanding job location, sectorial concentration, and transit use among low-wage
workers. Journal of Transport Geography, 52, 1 - 11.
64
Levinson, D. (1998). Accessibility and the journey to work Journal of Transport Geography, 6(1),
11-21.
Levinson, D. (2014). Identifying winners and losers in transportation. Transportation Research
Record(1812), 179 - 185.
Levinson, D., Marion, B., Owen, A., & Cui, M. (2017). The City is flatter: Changing patterns of job
and labor access. Cities, 60, 124 - 138.
Litman, T. (2002). Evaluating transportation equity. World Transport Policy & Practice, 8(2), 50 -
65.
Lucas, K. (2012). Transport and social exclusion: Where are we now? Transport Policy, 20, 105 -
113.
Lucas, K., van Wee, B., & Maat, K. (2016). A method to evaluate equitable accessibility:
Combining ethical theories and accessibility-based approaches. Transportation, 43(3), 473
- 490.
Luo, W., & Wang, F. (2003). Measures of spatial accessibility to health care in a GIS environment:
synthesis and a case study in the Chicago region. Environment and Planning B: Planning
and Design, 30, 865 - 884.
Lyons, M. (1996). Gentrification, socioeconomic change, and the geography of displacement.
Journal of Urban Affairs, 18(1), 39 - 62.
Manaugh, K., & El-Geneidy, A. (2012). Who benefits from new transportation infrastructure? Using
accessibility measures to evaluate social equity in transit provision. In K. Geurs, K. Krizek,
& A. Reggiani (Eds.), Accessibility and Transport Planning: Challenges for Europe and
North America. London, UK: Edward Elgar.
Mao, L., & Nekorchuk, D. (2013). Measuring spatial accessibility to healthcare for populations with
multiple transportation modes. Health & Place, 24, 115 - 122.
Martínez, L., & Viegas, J. (2009). Effects of transportation accessibility on residential property
values. Transportation Research Record(2115), 127 - 137.
Mathur, S., & Ferrell, C. (2013). Measuring the impact of sub-urban transit-oriented developments
on single-family home values. Transportation Research Part A, 47, 42 - 55.
Mayor's Council on Regional Transportation. (2014). Regional transportation investments: A vision
for Metro Vancouver.
Medda, F. (2012). Land value capture finance for transport accessibility: A review. Journal of
Transport Geography, 25, 154 - 161.
65
Mejia-Dorantes, L., Paez, A., & Vassallo, J. (2012). Transportation infrastructure impacts on firm
location: The effect of a new metro line in the suburbs of Madrid. Journal of Transport
Geography, 22, 236 - 250.
Metrolinx. (2008). The Big Move.
Metrolinx. (2016). Discussion paper for the next regional transportation plan.
Mills, E. (1967). An aggregative model of resource allocation in a metropolitan area. The American
Economic Review, 57(2), 197 - 210.
Morris, J. M., Dumble, P. L., & Wigan, M. R. (1979). Accessibility indicators for transport planning.
Transportation Research Part A, 13(2), 91 - 109.
Mumford, L. (1968). Neighborhood and neighborhood unit The urban prospect. New York:
Harcourt, Brace and World.
Newman, P. (2009). Planning for transit oriented development: Strategic principles. In C. Curtis,
J. Renne, & L. Bertolini (Eds.), Transit Oriented Development: Making it Happen (pp. 13 -
22). Surrey: Ashgate Publishing Limited.
Newman, P. W. G., & Kenworthy, J. R. (1996). The land use - transport connection. Land Use
Policy, 13(1), 1 - 22.
Niemeier, D. (1997). Accessibility: An evaluation using consumer welfare. Transportation, 24(4),
377 - 396.
North Central Texas Council of Governments. (2016). Mobility 2040.
NSW Government. (2012). NSW Long Term Transport Master Plan.
Owen, A., & Levinson, D. (2015a). Modeling the commute mode share of transit using continuous
access to jobs. Transportation Research Part A, 74, 110 - 122.
Owen, A., & Levinson, D. (2015b). Modeling the commute mode share of transit using continuous
accessibility to jobs. Transportation Research Part A, 74, 110 - 122.
Owen, A., Murphy, B., & Levinson, D. (2016). Access across America: Transit 2016. Retrieved
from http://ao.umn.edu/research/america/transit/2016/index.html
Owen, W. (1966). The metropolitan transportation problem. Washington, DC: Brookings Institution
Press.
Ozbay, K., Ozmen-Ertekin, D., & Berechman, J. (2003). Empirical analysis of relationship between
accessibility and economic development. Journal of Urban Planning and Development,
129(2), 97 - 119.
Padeiro, M. (2013). Transport infrastructures and employment growth in the Paris metropolitan
margins. Journal of Transport Geography, 31, 44 - 53.
66
Paez, A., Mercado, R., Farber, S., Morency, C., & Roorda, M. (2010). Accessibility to health care
facilities in Montreal Island: an application of relative accessibility indicators from the
perspective of senior and non-senior residents. International Journal of Health
Geographics, 9(52), 1 - 15.
Papa, E., & Bertolini, L. (2015). Accessibility and Transit-Oriented Development in European
metropolitan areas. Journal of Transport Geography, 47, 70 - 83.
Pollack, S., Bluestone, B., & Billingham, C. (2011). Demographic change, diversity and
displacement in newly transit-rich neighborhoods. Paper presented at the Transportation
research board 90th annual meeting.
Potoglou, D., & Kanaroglou, P. (2008). Modelling car ownership in urban areas: a case study of
Hamilton, Canada. Journal of Transport Geography, 16, 42 - 54.
Proffitt, D., Bartholomew, K., Ewing, R., & Miller, H. (2017). Accessibility planning in American
metropolitan areas: Are we there yet? Urban Studies.
Proffitt, D. G., Bartholomew, K., Ewing, R., & Miller, H. (2015). Accessibility planning in american
metropolitan areas: Are we there yet? Paper presented at the Transportation Research
Board 94th Annual meeting, Washington, D.C.
Proffitt, D. G., Bartholomew, K., Ewing, R., & Miller, H. J. (2017). Accessibility planning in American
metropolitan areas: Are we there yet? Urban Studies, 1 - 26.
Putnam, S. H. (1983). Integrated urban models. London: Pion.
Ramsey, K., & Bell, A. (2014). The smart location database: A nationwide data resource
characterizing the built environment and destination accessibility at the neighborhood
scale. Cityscape: A Journal of Policy Development and Research, 16(2), 145 - 162.
Ratner, K., & Goetz, A. (2013). The reshaping of land use and urban form in Denver through
transit-oriented development. Cities, 30, 31 - 46.
Renne, J. (2009). From transit-adjacent to transit-oriented development. Local Environment,
14(1), 1 - 15.
Salon, D., & Shewmake, S. (2014). Opportunities for value capture to fund public transport: A
comprehensive review of the literature with a focus on East Asia. Retrieved from
https://www.itdp.org/wp-
content/uploads/2014/07/Salon_and_Shewmake_Opportunities_for_Value_Capture_to_
Fund_Public_Transport.pdf
San Diego Association of Governments. (2011). 2050 Regional Transportation Plan.
67
Shen, Q. (1998). Location characteristics of inner-city neighborhoods and employment
accessibility of low-wage workers. Environment and Planning B: Planning and Design, 25,
345 - 365.
Shen, Y., de Abreu e Silva, J., & Martínez, L. M. (2014). Assessing land cover change in large
urban areas resulting from High-speed Rail: A case study of Madrid Atocha railway station
from 1990 to 2006. Journal of Transport Geography, 41, 184 - 196.
Smolka, M. (2013). Implementing value capture in Latin America: Policies and tools for urban
development. Retrieved from Cambridge, MA:
http://www.lincolninst.edu/sites/default/files/pubfiles/implementing-value-capture-in-latin-
america-full_1.pdf
Southern California Association of Governments. (2016). Regional Transportation Plan 2040.
Stanley, J., & Lucas, K. (2008). Social exclusion: What can public transport offer? Research in
transportation economics, 22(1), 36-40.
Staricco, L., & Vitale Brovarone, E. (2018). Promoting TOD through regional planning. A
comparative analysis of two European approaches. Journal of Transport Geography, 66,
45 - 52.
Statistics Canada. (2010). Mode of transportation and average commuting time to get to work in
Montréal, Toronto and Vancouver census metropolitan areas. Retrieved from
http://www.statcan.gc.ca/pub/11-008-x/2011002/t/11531/tblbx-eng.htm
Statistics Canada. (2015). Labour force survey estimates (LFS), employment by census
metropolitan area - CANSIM table 282-0130. Retrieved from
http://www5.statcan.gc.ca/cansim/a26?lang=eng&retrLang=eng&id=2820130&&pattern=
&stByVal=1&p1=1&p2=31&tabMode=dataTable&csid=
Sultana, S. (2017). Transportation and land use. In D. Richardson (Ed.), The International
Encyclopedia of Geography.
The City of Calgary. (2009). Calgary transportation plan.
The Spectator. (1933, 26 May). A changing metropolis. The Spectator.
Tomer, A., Kneebone, E., Puentes, R., & Berube, A. (2011). Missed opportunity: Transit and jobs
in metropolitan America. Retrieved from https://www.brookings.edu/wp-
content/uploads/2016/06/0512_jobs_transit.pdf
Transport for London. (2006). Tranpsort 2025: Transport Vision for a Growing World City.
Transport for London. (2017). Land value capture. Retrieved from London:
https://www.london.gov.uk/sites/default/files/land_value_capture_report_transport_for_lo
ndon.pdf
68
Tyndall, J. (2015). Waiting for the R train: Public transportation and employment. Urban Studies,
54(2), 520 - 537.
van Wee, B., & Geurs, K. (2011). Discussing equity and social exclusion in accessibility
evaluations. European Journal of Transport and Infrastructure Research, 11(4).
Ville de Montréal. (2008). Montreal Transportation Plan.
Warade, R. (2007). The accessibility and development impacts of new transit infrastructure: The
circle line in Chicago. (Master in City Planning), Massachusetts Institute of Technology,
Boston, MA.
Warner, S. B. (1962). Streetcar suburbs: The process of growth in Boston, 1870-1900. Cambridge:
Harvard University Press.
Wickstrom, G. (1971). Defining balanced transportation: A question of opprtunity. Traffic Quarterly,
25(3), 337-349.
Zhao, Z., Iacono, M., Lari, A., & Levinson, D. (2012). Value capture for transportation finance.
Procedia - Social and Behavioral Sciences, 48, 435 - 448.
69